# Claude Blog - llms-full.txt # https://claude-blog.md/ ## Use the hosted skill vault - [Browse the skill vault](https://claude-blog.md/mcp): read the skill files without installing them. - [Markdown skill catalog](https://claude-blog.md/index.md): short links to all 32 skills. - [Agent procedure](https://claude-blog.md/mcp/AGENTS.md): discover, inspect, load an exact version, then read needed references. - [Source and attribution](https://claude-blog.md/mcp/SOURCE.md): pinned upstream files and hosting boundaries. - [Trust and capabilities](https://claude-blog.md/mcp/TRUST.md): task-scoped guidance, citation checks and execution limits. - [MCP endpoint](https://claude-blog.md/mcp): compatible clients connect here using protocol POST requests. - [MCP Server Card](https://claude-blog.md/mcp/server-card): connection details for the MCP endpoint (SEP-2127 draft format). - [Agent resource catalog](https://claude-blog.md/.well-known/ard.json): Agentic Resource Discovery catalog of the MCP server and all 32 skills. Also served at /.well-known/ai-catalog.json. The hosted vault provides instructions and declared references. It does not install tools or provide the full installed execution runtime. A user must provide the URL or configure the MCP connection. Browser WebMCP tools are available only in supporting clients. Availability does not guarantee automatic discovery or successful execution by every agent. > Full text of every sitemapped page on claude-blog.md for single-fetch AI ingestion. > Summary index lives at /llms.txt. Entries are rebuilt from current static HTML. # Generated: 2026-09-19 --- ## Claude Blog - Free AI Blog Writing Tool for Claude Code URL: https://claude-blog.md/ Claude Blog - Free AI Blog Writing Tool for Claude Code Skip to main content CLAUDE BLOG DEMO FEATURES COMMANDS SKILLS SKILL VAULT BLOG FAQ DEMOFEATURESCOMMANDSSKILLSSKILL VAULTBLOGFAQ v2.2.0 // FREE // OPEN SOURCE // MIT YOUR CONTENT TEAM IN THE TERMINAL. 31 sub-skills. 5 AI agents. 12 content templates. One install. Claude Blog helps create, review, and optimize useful, source-backed articles for readers, Google Search, and AI answer surfaces. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY WORKS ON MACOS, LINUX, WINDOWS. WINDOWS POWERSHELL INSTALL Prefer to use your agent without installing skill files? Open the skill vault or read the Markdown catalog. ★ 1,800+ STARS V2.2.0 NOTES SEE DEMO FEATURES DOCS // SEE IT IN ACTION AI BLOG WRITING IN 10 MINUTES THE FULL DEMO. TYPE ONE COMMAND. 4 AI AGENTS RESEARCH, WRITE, CHECK, AND SCORE YOUR POST. 14,100+ VIEWS 275 LIKES 10:39 RUNTIME BY AGRICI DANIEL WATCH ON YOUTUBE → Blog commands in action The /blog write command generating a post // INSTALL CLONE. INSTALL. WRITE. 01 Clone git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git 02 Install bash claude-blog/install.sh Windows: powershell -ExecutionPolicy Bypass -File claude-blog\install.ps1 03 Write claude /blog write "your topic here" 4 AI agents research, draft, check, and score your post. 31 SUB-SKILLS 5 AI AGENTS 30 COMMANDS 12 TEMPLATES $0 FOREVER. MIT LICENSED. 1,800+ GITHUB STARS 300+ FORKS 14,100+ YOUTUBE VIEWS 275 YOUTUBE LIKES // FEATURES WHAT IT DOES 01 4-AGENT PIPELINE [4 AI AGENTS] RESEARCHER FINDS DATA AND STATISTICS. WRITER DRAFTS THE ARTICLE. CHECKER VALIDATES SEO AND READABILITY. REVIEWER SCORES AND POLISHES. FOUR SPECIALIZED AGENTS WORK IN SEQUENCE ON EVERY POST. 02 12 CONTENT TEMPLATES [TEMPLATES] HOW-TO GUIDE, LISTICLE, CASE STUDY, COMPARISON, PILLAR PAGE, PRODUCT REVIEW, THOUGHT LEADERSHIP, ROUNDUP, TUTORIAL, NEWS ANALYSIS, DATA RESEARCH, FAQ KNOWLEDGE BASE. 03 SEARCH + AI READINESS [GOOGLE + AI] PEOPLE-FIRST PUBLISHING WITH CURRENT GOOGLE GUIDANCE, SOURCE FIDELITY, CRAWLABILITY, AND AN INTERNAL AI CITATION READINESS HEURISTIC. NO GUARANTEED RANKINGS OR CITATIONS. 04 100-POINT SCORING [EDITORIAL HEURISTIC] ORIGINALITY, PURPOSE FIT, SOURCE FIDELITY, FACTUAL SUPPORT, VISIBLE-CONTENT CONSISTENCY, CRAWLABILITY, MEDIA RELEVANCE, AND READER USEFULNESS. 05 CLAUDE SEO INTEGRATION [ECOSYSTEM] CONSUMES SEO AUDIT DATA FROM CLAUDE SEO. KEYWORD GAPS, CONTENT OPPORTUNITIES, AND TECHNICAL REQUIREMENTS FLOW INTO YOUR BLOG PLANNING. 06 EDITORIAL WORKFLOW [LIFECYCLE] CALENDAR, OUTLINE, BRIEF, WRITE, SCORE, OPTIMIZE, REPURPOSE. FULL CONTENT LIFECYCLE MANAGEMENT IN YOUR TERMINAL. CONTENT CALENDAR KEYWORD CANNIBALIZATION FACT CHECKING WRITING PERSONAS CMS TAXONOMY AI IMAGE GENERATION CONTENT REPURPOSING SVG CHARTS SCHEMA MARKUP // COMMANDS EVERY COMMAND /blog write WRITE A NEW BLOG POST FROM SCRATCH /blog rewrite OPTIMIZE AN EXISTING BLOG POST /blog analyze QUALITY AUDIT WITH 0-100 SCORE /blog brief GENERATE A DETAILED CONTENT BRIEF /blog calendarGENERATE AN EDITORIAL CALENDAR /blog strategy BLOG STRATEGY AND TOPIC IDEATION /blog outline SERP-INFORMED CONTENT OUTLINE /blog seo-check POST-WRITING SEO VALIDATION /blog schema GENERATE JSON-LD SCHEMA MARKUP /blog repurpose REPURPOSE FOR SOCIAL, EMAIL, YOUTUBE /blog geo AI CITATION READINESS AUDIT /blog imageAI IMAGE GENERATION VIA GEMINI /blog audit [dir]FULL-SITE BLOG HEALTH ASSESSMENT /blog cannibalization [dir]DETECT KEYWORD OVERLAP /blog factcheck VERIFY STATISTICS AGAINST SOURCES /blog personaMANAGE WRITING PERSONAS AND VOICE /blog taxonomyTAG AND CATEGORY CMS MANAGEMENT /blog flow [stage] FLOW FRAMEWORK: 30 EVIDENCE-LED PROMPTS /blog google [cmd] GOOGLE APIS: GSC, PSI, CRUX, GA4, INDEXING /blog notebooklm SOURCE-GROUNDED RESEARCH VIA NOTEBOOKLM /blog audio generate AI AUDIO NARRATION: 30 VOICES, 80+ LANGUAGES /blog cluster plan SERP-OVERLAP TOPIC CLUSTERS [COMMUNITY] /blog multilingual WRITE + TRANSLATE + LOCALIZE + HREFLANG [COMMUNITY] /blog translate SEO TRANSLATION WITH FORMAT PRESERVATION [COMMUNITY] /blog localize CULTURAL DEEP-ADAPTATION PER LOCALE [COMMUNITY] /blog locale-audit MULTILINGUAL QA: COMPLETENESS + HREFLANG [COMMUNITY] /blog brand [init|show|update]DURABLE BRAND.md + VOICE.md CONTEXT /blog chartINLINE SVG DATA CHARTS: 7 TYPES, DARK-MODE /blog discourse API-FREE 30-DAY DISCOURSE RESEARCH // HOW SCORING WORKS 0-100 QUALITY SCORE CONTENT QUALITY 30% SEO OPTIMIZATION 25% E-E-A-T SIGNALS 15% TECHNICAL ELEMENTS 15% AI READINESS HEURISTIC 15% CRITICAL UNSUPPORTED FACTUAL CLAIMS, MISSING PRIMARY CONTENT, SCHEMA THAT CONTRADICTS THE PAGE HIGH WEAK SOURCE FIDELITY, COMMODITY SUMMARY, INACCESSIBLE KEY MATERIAL, MISLEADING EXPERIENCE CLAIMS MEDIUM CRAWLABILITY, CANONICAL, MEDIA RELEVANCE, OR VISIBLE-CONTENT CONSISTENCY NEEDS WORK STYLE PROJECT PROSE RULES AND OPTIONAL PRESENTATION IMPROVEMENTS, NOT GOOGLE REQUIREMENTS // RESULTS ILLUSTRATIVE SCORES Example editorial findings, not calibrated ranking or citation predictions. 87/100 How to Write SEO Blog Posts with AI TUTORIAL // 2,000 WORDS Content Quality: 28/30 - strong depth, clear structure SEO: 23/25 - keyword placement, heading hierarchy E-E-A-T: 12/15 - needs stronger source support AI readiness: 14/15 - clear entities and sourced claims 74/100 Claude Blog vs Jasper vs Writesonic COMPARISON // 1,800 WORDS Content Quality: 24/30 - good depth on features SEO: 19/25 - title tag could be tighter E-E-A-T: 10/15 - document methodology before claiming a test Technical: 11/15 - visible comparison and schema need alignment 91/100 How to Optimize for Google AND AI Citations GUIDE // 1,800 WORDS Content Quality: 29/30 - comprehensive, original framework SEO: 24/25 - strong keyword integration E-E-A-T: 13/15 - includes practical examples AI readiness: 15/15 - useful structure and source alignment // COMMUNITY WHAT THEY SAID Real comments from the YouTube demo video. "I've been using your ads and your seo skills, going to try this one out now. ALL OPEN SOURCE YOU ARE INSANE" YOUTUBE COMMENTER // YOUTUBE "I have used a similar method to write blog posts, it took me way longer and the result was worse. Thank you so much for sharing it and improving my blogs!" YOUTUBE COMMENTER // YOUTUBE "This is crazy good. Awesome work. Question: I currently use SEO Surfer and their human writer / AI detector. How good is the produced text from the writer skill?" YOUTUBE COMMENTER // YOUTUBE VIEW ALL COMMENTS ON YOUTUBE → // FAQ QUESTIONS PEOPLE ACTUALLY ASK WHAT IS CLAUDE BLOG? - Claude Blog is a free, open-source Claude Code toolkit with 31 sub-skills, 5 specialized AI agents, and 12 content templates. It creates, reviews, and optimizes useful source-backed articles with evidence-led Google guidance and an internal AI citation readiness heuristic. MIT licensed with more than 1,800 GitHub stars. HOW DO I INSTALL CLAUDE BLOG? + On Unix, macOS, or Linux: git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git then bash claude-blog/install.sh. On Windows: git clone then powershell -ExecutionPolicy Bypass -File claude-blog\install.ps1. Requires Python 3.11+ and Claude Code CLI. IS CLAUDE BLOG FREE? + Yes. Claude Blog is fully open source under the MIT license with no paid tiers. The core tool runs entirely through Claude Code. Optional extensions like Banana (AI image generation via Google Gemini) require a free API key from Google. WHAT CONTENT TEMPLATES ARE AVAILABLE? + Claude Blog includes 12 content templates: how-to guide, listicle, case study, comparison, pillar page, product review, thought leadership, roundup, tutorial, news analysis, data research, and FAQ knowledge base. Each is a starting structure that should be adapted to reader intent. There is no Google word-count target or mandatory FAQ. HOW DOES THE 100-POINT SCORING SYSTEM WORK? + Every blog post receives an internal 0 to 100 editorial heuristic across five compatible categories. The weights are not direct ranking factors, and the AI readiness component is not a calibrated citation probability. The quality gate remains 70. DOES CLAUDE BLOG WORK WITH CLAUDE SEO? + Yes. Claude Blog integrates with Claude SEO. SEO audit findings like keyword gaps, content opportunities, and technical requirements flow directly into blog planning and content strategy. Both tools are free and open source. CAN IT GENERATE CONTENT FOR ANY NICHE OR PLATFORM? + Yes. Claude Blog generates content for any topic and supports multiple platforms including Next.js/MDX, Astro, Hugo, Jekyll, WordPress, Ghost, 11ty, Gatsby, and static HTML. Writing personas let you customize tone and voice per brand. HOW DOES IT ASSESS AI CITATION READINESS? + The /blog geo command reviews source fidelity, factual support, entity clarity, accessible structure, crawlability, and useful non-commodity value. It does not guarantee citation by Google, ChatGPT, Perplexity, Bing, or any other system. // ECOSYSTEM WORKS WITH THESE CLAUDE BLOG THIS PROJECT. 31 SUB-SKILLS, 5 AI AGENTS, 30 COMMANDS, 12 CONTENT TEMPLATES. EVIDENCE-LED GOOGLE GUIDANCE AND AI READINESS HEURISTICS. FREE FOREVER. AGRICIDANIEL/CLAUDE-BLOG → CLAUDE SEO 25 SUB-SKILLS, 18 PARALLEL AI AGENTS. INTEGRATES WITH AHREFS AND SEMRUSH MCP. FREE FOREVER. CLAUDE-SEO.MD → CLAUDE BANANA AI IMAGE GENERATION VIA GOOGLE GEMINI. GENERATES OG IMAGES, HERO GRAPHICS, AND BLOG ASSETS FROM TEXT PROMPTS. AGRICIDANIEL/BANANA-CLAUDE → [THE PITCH] YOU ALREADY HAVE A TERMINAL OPEN. 1,800+ GitHub stars. 14,100+ YouTube views. MIT licensed, free forever. The install takes 30 seconds. The write command handles the rest. You review the evidence and final draft. COPY THE INSTALL COMMANDGITHUB > ABOUT CLAUDE BLOG Claude Blog is a free, open-source AI blog writing tool built for Claude Code. It is an alternative to paid content platforms like Jasper, Copy.ai, and Writesonic. Instead of a browser-based editor, Claude Blog runs in your terminal. You type /blog write and 4 AI agents research, draft, check, and score your blog post, producing a 0-100 quality score with actionable improvements. The tool covers writing, rewriting, editorial analysis, briefs, calendars, strategy, outlines, source verification, schema consistency, repurposing, AI citation readiness, image generation, site audits, cannibalization, personas, taxonomy, brand context, charts, discourse research, multilingual publishing, audio, Google data, and content-decay review. Claude Blog supports 12 content templates that adapt to reader intent. v2.2.0 adds verified Google currentness, a full pull-request and issue audit, analyzer fixes, safer installers, and a market-ready source-cited Brain without promising rankings or citations. MIT licensed. More than 1,800 GitHub stars and 300 forks as verified on August 26, 2026. Works on macOS, Linux, and Windows. Requires Python 3.11+ and Claude Code CLI. Created by AgriciDaniel. Watch the demo on YouTube. Join the community on Skool. Connect on LinkedIn. CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL AGRICIDANIEL.COM BLOG WRITER DEEP DIVE CLAUDE SEO RANKENSTEIN GITHUB SKILLS BLOG YOUTUBE SKOOL LINKEDIN DOCS CLAUDE-BLOG.MD LLMS.TXT CHANGELOG LICENSE --- ## All 30 Blog Skills - Free AI Tools | Claude Blog URL: https://claude-blog.md/skills All 30 Blog Skills - Free AI Tools | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills // ALL SKILLS 31 BLOG SUB-SKILLS. ONE INSTALL. Use this skill in the hosted vault Every blog workflow is covered by a specialized skill. Each can run independently or as part of a full content pipeline. Write, score, optimize, repurpose, translate, and publish. Version 2.2.0 adds verified Google currentness, analyzer fixes, safer installers, and a market-ready source-cited Brain. 31 SUB-SKILLS 5 AI AGENTS PER POST 30 COMMANDS $0 FOREVER. MIT LICENSED. 01 AI BLOG WRITER 4 AI agents research, write, check, and score your post from a single command. /blog write LEARN MORE → 02 CONTENT SCORING 5-category 100-point editorial heuristic with source, usefulness, and readability checks. /blog analyze LEARN MORE → 03 CONTENT OPTIMIZER Rewrite and optimize existing blog posts for rankings and AI citations. /blog rewrite LEARN MORE → 04 BLOG STRATEGY Topic cluster architecture, audience mapping, and competitive landscape analysis. /blog strategy LEARN MORE → 05 CONTENT BRIEFS Detailed briefs with target keywords, outlines, and competitive analysis. /blog brief LEARN MORE → 06 EDITORIAL CALENDAR Monthly content calendars with topic clusters and publishing schedules. /blog calendar LEARN MORE → 07 CONTENT OUTLINES SERP-informed outlines with logical headings and intent-dependent coverage. /blog outline LEARN MORE → 08 SEO VALIDATION Post-writing SEO checklist: title, meta, headings, links, schema, OG tags. /blog seo-check LEARN MORE → 09 SCHEMA GENERATOR JSON-LD schema markup for BlogPosting, FAQPage, Person, and Breadcrumbs. /blog schema LEARN MORE → 10 CONTENT REPURPOSING Transform posts into Twitter threads, LinkedIn articles, YouTube scripts, and email. /blog repurpose LEARN MORE → 11 AI CITATION AUDIT Optimize for Google AI Overviews, ChatGPT, Perplexity, and Bing Copilot. /blog geo LEARN MORE → 12 AI IMAGE GENERATION Generate blog hero images and OG graphics via Google Gemini API. /blog image LEARN MORE → 13 SITE BLOG AUDIT Full-site blog health assessment with per-post scores and action queue. /blog audit [dir] LEARN MORE → 14 KEYWORD OVERLAP Detect posts competing for the same search intent and keywords. /blog cannibalization [dir] LEARN MORE → 15 FACT CHECKING Verify statistics and claims against cited source URLs. /blog factcheck LEARN MORE → 16 WRITING PERSONAS Create and manage tone profiles with 4-dimension voice framework. /blog persona LEARN MORE → 17 CMS TAXONOMY Tag and category management for WordPress, Shopify, Ghost, Strapi, Sanity. /blog taxonomy LEARN MORE → // NEW IN V1.6.5 - V2.2.0 FLAGSHIP & COMMUNITY SKILLS The ecosystem now includes 31 sub-skills. Five are community contributions from the Pro Hub Challenge. Version 2.2.0 adds executable currentness checks, a full backlog audit, source-review records, and stronger release validation. 18 FLOW FRAMEWORK Evidence-led blogging. 30 AI prompts across Find, Optimize, Win stages. CC BY 4.0. Context-matched orchestration. /blog flow [stage] LEARN MORE → 19 GOOGLE APIS Search Console, PageSpeed, CrUX 25-week history, Indexing, GA4, YouTube, NLP, Keywords. 13 commands, 4 credential tiers. /blog google [cmd] LEARN MORE → 20 NOTEBOOKLM RESEARCH Source-grounded research via Google NotebookLM. Query your Drive notebooks for citation-ready answers with exact file-and-page attribution. /blog notebooklm LEARN MORE → 21 AUDIO NARRATION AI audio narration via Gemini TTS. 30 voices, 80+ languages, multi-speaker dialogue, 24kHz mono WAV output. /blog audio generate LEARN MORE → 22 TOPIC CLUSTERS SERP-based content clusters with hub-and-spoke architecture and automatic internal-link injection. Pro Hub Challenge winner: Lutfiya Miller. /blog cluster plan LEARN MORE → 23 MULTILINGUAL PUBLISHING One command: write + translate + localize + hreflang across multiple languages. Pro Hub Challenge: Chris Mueller. /blog multilingual LEARN MORE → 24 SEO TRANSLATION Native-quality SEO translation with format preservation. Markdown, frontmatter, schema, SVG, locale-correct formatting. Community: Chris Mueller. /blog translate LEARN MORE → 25 CULTURAL LOCALIZATION Deep cultural adaptation: idioms, examples, references per locale. DACH, Francophone, Hispanic, Japanese profiles. Community: Chris Mueller. /blog localize LEARN MORE → 26 MULTILINGUAL AUDIT QA across all language variants: completeness, hreflang correctness, meta parity, freshness tracking. Community: Chris Mueller. /blog locale-audit LEARN MORE → 27 BRAND CONTEXT Durable BRAND.md and VOICE.md, auto-loaded by every sub-skill. Audience, positioning, voice, and editorial rules from one source. New in v1.8.0. /blog brand init LEARN MORE → 28 SVG DATA CHARTS Dark-mode inline SVG charts: bar, donut, line, lollipop, area, radar. Accessible and source-attributed. Auto-runs inside write and rewrite. New in v1.8.0. /blog chart LEARN MORE → 29 DISCOURSE RESEARCH API-free 30-day research across Reddit, X, YouTube, and Hacker News. A voice-of-customer brief the writer consumes. New in v1.8.0. /blog discourse LEARN MORE → FIVE COMMUNITY-CONTRIBUTED SKILLS · VIEW CONTRIBUTORS INSTALL ALL 30 SKILLS IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE SKOOL LINKEDIN LLMS.TXT --- ## Blog - AI Content Creation Guides | Claude Blog URL: https://claude-blog.md/blog Blog - AI Content Creation Guides | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog // BLOG GUIDES AND TUTORIALS. Practical guides on AI-powered blog writing, content scoring, template selection, and optimizing for both Google and AI citation platforms. BUILDING IN PUBLIC I Turned Claude Blog Into a Web Skill Vault A Markdown-first skill vault, a live video demo, and what agents can do with a link. SEPTEMBER 19, 2026 · DANIEL AGRICIRead the experiment → FACT-CHECKING How to Fact-Check AI-Written Blog Posts: A Claim-by-Claim Workflow A practical workflow for checking the numbers, quotes, policies, and product claims in AI-written drafts before publication. DANIEL AGRICI // SEPTEMBER 7, 2026 // 5 MIN READ READ ARTICLE → RELEASE UPDATE Claude Blog v2.2.0: Verified Google Guidance and a Market-Ready Brain A skeptical release audit turned every pull request, issue, Google update, dependency, installer, and Brain source into a claim that had to earn its place. DANIEL AGRICI // AUGUST 25, 2026 // 3 MIN READ READ ARTICLE → RELEASE UPDATE Claude Blog v2.1.1: Google Guidance, Honest Scoring, Safer Releases The public v2.1.1 release replaces rigid SEO folklore with evidence-led editorial heuristics, updates technical Search guidance, and makes every release surface auditable. DANIEL AGRICI // JULY 23, 2026 // 3 MIN READ READ ARTICLE → DEEP DIVE Claude Watermarks + SynthID: A Marketer's Field Guide Claude now watermarks its text using DeepMind's SynthID approach, and the EU AI Act made machine-readable marking mandatory. What survives editing, what platforms label, and how content workflows should adapt. DANIEL AGRICI // AUG 20, 2026 // 9 MIN READ READ ARTICLE → STRATEGY Public Questions and PAA SEO: Turn Search Demand Into AI Answers Seed terms carry the demand. People Also Ask questions carry the public language. This source-backed workflow turns both into answer-first pages for Google and AI search. DANIEL AGRICI // MAY 28, 2026 // 7 MIN READ READ ARTICLE → RELEASE UPDATE Claude Obsidian v1.9: Your AI Second Brain Gets a Compound Memory The big claude-obsidian update. v1.9 Compound Vault brings hybrid retrieval, methodology modes, a thinking framework, and multi-writer safety to your AI second brain. Free and MIT licensed, still plain Markdown you own. DANIEL AGRICI // MAY 28, 2026 // 6 MIN READ READ ARTICLE → RELEASE UPDATE Claude Blog v1.9.1: 5 Gates Block Sloppy Drafts + 95/100 Audit v1.9.0 introduced a 5-gate Blog Delivery Contract that blocks drafts under 90 of 100. v1.9.1 closed 11 audit findings, all HIGH and MEDIUM. 187 tests, 142 hash-pinned packages, 8-agent cybersecurity audit pass. DANIEL AGRICI // MAY 21, 2026 // 9 MIN READ READ ARTICLE → COMPARISON Claude Code vs Codex: Who Should Review Whom? An evidence-first cross-model workflow for choosing the writer, independent reviewer, review scope, quality gates, and stopping rule. DANIEL AGRICI // UPDATED AUG 4, 2026 // 10 MIN READ READ ARTICLE → RELEASE UPDATE Claude Blog v1.7.0: 5 Community Skills Pro Hub Release Pro Hub Challenge recap. Lutfiya Miller's semantic cluster engine and Chris Mueller's multilingual suite (4 skills) integrated into Claude Blog v1.7.0. Plus FLOW framework. DANIEL AGRICI // MAY 11, 2026 // 7 MIN READ READ ARTICLE → SECURITY Claude Blog v1.7.1: Security Hardening Arc v1.7.1 closes 1 CRITICAL plus 5 HIGH vulnerabilities. VULN-001 API key exposure fixed, OAuth CSRF protection, hash-pinned dependencies, PowerShell 5.1 compatibility. DANIEL AGRICI // MAY 11, 2026 // 8 MIN READ READ ARTICLE → DEEP DIVE Multilingual Blog Publishing: One Command, Every Locale One command, four languages. Translation with format preservation, cultural deep-adaptation, hreflang generation, multilingual QA. Free with Claude Blog. DANIEL AGRICI // MAY 11, 2026 // 8 MIN READ READ ARTICLE → PLUGIN I Turned Obsidian Into a Self-Organizing AI Second Brain claude-obsidian turns Obsidian into a self-organizing AI second brain. 15 skills, plain Markdown you own, zero manual filing. A free, MIT-licensed Claude Code plugin built on Karpathy's LLM Wiki pattern. DANIEL AGRICI // MAY 28, 2026 // 10 MIN READ READ ARTICLE → CHANGELOG claude-blog v1.6.6-v1.6.8: 29 Accuracy Fixes and What Changed Two independent audits, 88 items reviewed, 29 corrections applied. GA4 quotas, AI crawler specs, schema deprecations, and the Windows PowerShell installer fix explained. DANIEL AGRICI // APRIL 8, 2026 // 8 MIN READ READ ARTICLE → TUTORIAL Publish Blog Posts to WordPress with AI (Free Plugin) Write with Claude Blog, publish to WordPress with WP MCP Ultimate. One conversation, zero copy-paste. 58 WordPress abilities, 2-minute setup. DANIEL AGRICI // MARCH 29, 2026 // 8 MIN READ READ ARTICLE → TUTORIAL How to Get Google SEO Data for Free (API Guide) Access Search Console, Core Web Vitals, and PageSpeed data from your terminal. $0.0006/query vs $500+/mo for Ahrefs. Free, open-source. DANIEL AGRICI // MARCH 28, 2026 // 8 MIN READ READ ARTICLE → TUTORIAL How to Write SEO Blog Posts with AI (Free Tool) Step-by-step guide to writing SEO blog posts with Claude Blog. Install, configure, and generate your first post with 4 AI agents. Free, open-source. DANIEL AGRICI // MARCH 25, 2026 // 8 MIN READ READ ARTICLE → COMPARISON Claude Blog vs Jasper vs Writesonic (2026) Free AI blog writer vs $49/mo Jasper and $16/mo Writesonic. Feature comparison, pricing, output quality, and who should use each tool. DANIEL AGRICI // MARCH 25, 2026 // 7 MIN READ READ ARTICLE → GUIDE 12 Blog Templates for Every Content Type Complete guide to Claude Blog's 12 content templates. How-to, listicle, case study, comparison, pillar page, and 7 more. When to use each template. DANIEL AGRICI // MARCH 25, 2026 // 6 MIN READ READ ARTICLE → GUIDE How AI Content Scoring Works (100-Point System) How Claude Blog's 100-point internal editorial heuristic works, what the five categories measure, and why the score is not a ranking prediction. DANIEL AGRICI // MARCH 25, 2026 // 5 MIN READ READ ARTICLE → CHECKLIST The Ultimate Blog SEO Checklist for 2026 The complete blog SEO checklist for 2026. Title tags, meta descriptions, heading hierarchy, internal links, schema markup, Core Web Vitals, and AI citations. DANIEL AGRICI // MARCH 25, 2026 // 6 MIN READ READ ARTICLE → GUIDE How to Optimize for Google AND AI Citations How to optimize blog posts for both Google rankings and AI citation platforms. Google AI Overviews, ChatGPT, Perplexity, and Bing Copilot. Free guide. DANIEL AGRICI // MARCH 25, 2026 // 7 MIN READ READ ARTICLE → LISTICLE Best Claude Code Skills in 2026: The Complete Guide The most impactful Claude Code skills in 2026. Blog writing, SEO audits, image generation, and more. Free, open-source tools for your terminal. DANIEL AGRICI // MARCH 25, 2026 // 9 MIN READ READ ARTICLE → TUTORIAL How to Build an AI Content Pipeline with Claude Blog Build an automated AI content pipeline from strategy to publishing. 7 steps using Claude Blog's terminal-based workflow. Free, open-source. DANIEL AGRICI // MARCH 25, 2026 // 8 MIN READ READ ARTICLE → COMPARISON The Best Free AI Blog Writing Tools in 2026 Free AI blog writing tools compared. Claude Blog, Copy.ai, BlogSEO.ai, ChatGPT, and Gemini. Features, limitations, and which to pick. DANIEL AGRICI // MARCH 25, 2026 // 7 MIN READ READ ARTICLE → TRY CLAUDE BLOG FOR YOURSELF. VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## Claude Blog · The Markdown vault URL: https://claude-blog.md/mcp Claude Blog · The Markdown vault Claude Blog Read as Markdown ↗Connect MCPHosted vault A small home for your next good idea Every blog skill. One open vault. Find the skill you need. Give your agent its Markdown. Go from a first thought to a finished piece, one useful instruction at a time. From the Claude Blog repository ↗ ✳ CLAUDE BLOG‹ Markdown ↗Connect MCP Expand allCollapse all Find your next step EXPLORERLoading Loading release…Hosted vaultRefresh Claude Blog Skill Vault Explore 32 blog skills for planning, writing, reviewing, and repurposing content. Read the Markdown instructions directly or connect a compatible agent through MCP. No skill-file installation is required; execution still depends on your agent’s tools and permissions. Browse the Markdown catalog · Read the agent guide · Explore the skill descriptions · Claude Blog home Opening the interactive explorer… For agents, too One connection. Skills on demand. Add this endpoint once in a compatible MCP client. Your agent can search, inspect and read the skills it needs. Or simply share a Markdown prompt, no individual skill installation required when your agent can fetch the link. Copy endpoint This demo serves instructions. Browsing, scripts, integrations and paid services depend on your agent’s tools and your authorization. Open the Markdown index ↗ Claude Blog · Markdown first. Open by design.Loading source revision… × Connect your agent Add this endpoint once in a compatible MCP client. Skills and references are retrieved on demand. Copy Or use any skill’s Markdown prompt. Your agent needs access to this URL. The vault serves instructions; execution depends on the host’s tools and permissions. Read the Markdown index ↗× Give your agent a starting point. Copy promptUse MCP instead --- ## AI Blog Writer - 4 Agent Pipeline | Claude Blog URL: https://claude-blog.md/skills/blog-write AI Blog Writer - 4 Agent Pipeline | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/AI Blog Writer // BLOG WRITE 4-AGENT AI BLOG WRITER Use this skill in the hosted vault Type /blog write and four specialized AI agents handle the rest. Researcher finds data and statistics. Writer drafts the article. Checker validates SEO and readability. Reviewer scores and polishes. One command, production-ready content. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog write COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS 4 AGENTS. ONE COMMAND. When you run /blog write, Claude Blog launches a 4-agent pipeline. Each agent specializes in one phase of content creation, and they run in sequence so each builds on the previous agent's output. The result is a publication-ready blog post with sourced statistics, proper SEO, and a quality score. 01 RESEARCH AGENT FINDS CURRENT STATISTICS, VERIFIES SOURCES AGAINST TIER 1-3 QUALITY STANDARDS, DISCOVERS IMAGES FROM PIXABAY AND UNSPLASH. 02 WRITER AGENT DRAFTS THE FULL ARTICLE WITH ANSWER-FIRST FORMATTING, PROPER HEADING HIERARCHY, AND SOURCED STATISTICS. 03 SEO CHECKER VALIDATES TITLE TAG, META DESCRIPTION, HEADING KEYWORDS, INTERNAL LINKS, AND SCHEMA MARKUP. 04 REVIEWER AGENT SCORES THE POST ON A 100-POINT SCALE ACROSS 5 CATEGORIES. IDENTIFIES WEAKNESSES AND SUGGESTS FIXES. 05 12 TEMPLATES HOW-TO, LISTICLE, CASE STUDY, COMPARISON, PILLAR PAGE, PRODUCT REVIEW, THOUGHT LEADERSHIP, ROUNDUP, TUTORIAL, NEWS, DATA, FAQ. 06 PLATFORM SUPPORT NEXT.JS/MDX, ASTRO, HUGO, JEKYLL, WORDPRESS, GHOST, 11TY, GATSBY, STATIC HTML. // USAGE HOW TO WRITE A POST Step 1: Install Claude Blog git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git bash claude-blog/install.sh Step 2: Open Claude Code claude Step 3: Write Your Post /blog write The pipeline asks for your topic and template preference, then runs all 4 agents in sequence. Runtime depends on research, evidence, media, complexity, and enabled checks. You get a reviewed, formatted post without padding it to a fixed length. // FAQ QUESTIONS ABOUT BLOG WRITE How does the 4-agent pipeline work? - The research agent gathers and verifies sources, the writer drafts useful content, the SEO checker reviews search and technical fundamentals, and the reviewer applies the internal editorial heuristic before delivery. Which content templates are available? + There are 12 adaptable starting structures, including how-to, case study, comparison, review, tutorial, news analysis, and data research. None imposes a Google word-count or FAQ requirement. Can I customize the writing voice? + Yes. Use /blog persona or project voice files to define a suitable tone while keeping claims and methodology accurate. What platforms does it support? + Next.js/MDX, Astro, Hugo, Jekyll, WordPress, Ghost, 11ty, Gatsby, and static HTML. Output format adapts to your platform. How long does it take to generate a post? + Runtime depends on research depth, source verification, media work, content complexity, and enabled checks. The workflow does not pad articles to a fixed length. // RELATED SKILLS EXPLORE MORE CONTENT SCORING Apply a 100-point internal editorial heuristic across content quality, search fundamentals, E-E-A-T signals, technical elements, and AI readiness. LEARN MORE → CONTENT OPTIMIZER Optimize existing blog posts for Google rankings and AI citations. Replace fabricated stats, add sourced data, improve readability. LEARN MORE → BLOG STRATEGY Plan your entire content strategy. Topic cluster architecture, audience mapping, competitive analysis, and editorial calendars. LEARN MORE → FLOW FRAMEWORK 30 evidence-led blogging prompts across Find, Optimize, Win stages. Structured prompts, not improvised queries. CC BY 4.0. FLOW FRAMEWORK → MULTILINGUAL PUBLISHING One command, every language. Source post plus translated and localized variants plus hreflang in a single pipeline. MULTILINGUAL PUBLISHING → SEO TRANSLATION Format-preserving translation: markdown, frontmatter, schema, locale-correct number and date formatting intact. SEO TRANSLATION → VIEW ALL SKILLS → WRITE YOUR FIRST POST IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE SKOOL LINKEDIN SKILLS LLMS.TXT CHANGELOG LICENSE --- ## Content Scoring - 100 Point System | Claude Blog URL: https://claude-blog.md/skills/blog-analyze Content Scoring - 100 Point System | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Content Scoring // BLOG ANALYZE 100-POINT CONTENT SCORING Use this skill in the hosted vault Every blog post gets a 100-point internal editorial score. Five weighted categories, readability context, and a prioritized improvement plan show what deserves review. It is not an AI detector or ranking prediction. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog analyze COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS 5 CATEGORIES. ONE SCORE. When you run /blog analyze, Claude Blog scores the post across 5 weighted editorial categories. It adds readability context and a prioritized list of improvements sorted by severity. TTR, burstiness, punctuation, and phrase patterns do not block publication by themselves. 01 5-CATEGORY SCORING CONTENT QUALITY (30), SEO OPTIMIZATION (25), E-E-A-T SIGNALS (15), TECHNICAL ELEMENTS (15), AI CITATION READINESS (15). 02 SOURCE & CLAIM REVIEW CHECKS SOURCE FIDELITY, FACTUAL SUPPORT, METHODOLOGY, ORIGINAL VALUE, AND UNSUPPORTED FIRST-HAND CLAIMS. 03 READABILITY FLESCH-KINCAID, GUNNING FOG, AND SMOG INDEX. SENTENCE LENGTH DISTRIBUTION AND PARAGRAPH STRUCTURE. 04 SCORING BANDS EXCEPTIONAL (90-100), STRONG (80-89), ACCEPTABLE (70-79), BELOW STANDARD (60-69), REWRITE NEEDED (UNDER 60). 05 ACTION PLAN PRIORITIZED FINDINGS: CRITICAL, HIGH, MEDIUM, LOW SEVERITY. SPECIFIC FIX SUGGESTIONS PER ISSUE. 06 BATCH ANALYSIS ANALYZE ENTIRE BLOG DIRECTORIES. PER-POST SCORES WITH SITE-WIDE HEALTH OVERVIEW. // SCORING CATEGORY WEIGHTS The 100-point score is a weighted aggregate. Content Quality carries the most weight because depth and readability matter most. AI Citation Readiness reflects the growing importance of GEO optimization. CONTENT QUALITY 30 SEO OPTIMIZATION 25 E-E-A-T SIGNALS 15 TECHNICAL 15 AI CITATION 15 Scoring Bands 90-100 Exceptional - publication-ready, best-in-class content 80-89 Strong - minor improvements possible, good to publish 70-79 Acceptable - some issues to address before publishing 60-69 Below Standard - significant improvements needed Under 60 Rewrite Needed - fundamental quality issues // USAGE HOW TO ANALYZE A POST Step 1: Install Claude Blog git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git bash claude-blog/install.sh Step 2: Open Claude Code claude Step 3: Analyze Your Post /blog analyze Point it at any blog post file or directory. You get a detailed heuristic breakdown, readability context, source and claim findings, and a prioritized action plan. // FAQ QUESTIONS ABOUT SCORING What does each scoring category measure? - Content Quality (30 pts) checks originality, purpose fit, usefulness, and clarity. SEO Optimization (25 pts) covers descriptive titles, logical headings, links, metadata, and crawlability. E-E-A-T Signals (15 pts) are editorial checks for source fidelity, factual support, authorship, and methodology. Technical Elements (15 pts) check schema consistency, relevant media, mobile, and social metadata. AI Citation Readiness (15 pts) is an internal heuristic for entity clarity, supported claims, and self-contained meaning. Is the score an AI detector or ranking prediction? + No. The score is an internal editorial heuristic, not an AI-authorship detector, calibrated citation probability, or Google ranking prediction. Readability and style signals provide context but do not block publication by themselves. Can I analyze posts from any platform? + Yes. The analyzer works with HTML, Markdown, MDX, and plain text files. It parses frontmatter, headings, paragraphs, images, links, and schema from any format. What is a good score? + The quality gate remains 70. Scores of 80 or higher indicate strong alignment with the internal editorial checks, but every finding still requires human judgment and source review. // RELATED SKILLS EXPLORE MORE AI BLOG WRITER Write SEO blog posts with a 4-agent pipeline. Research, draft, check, and review in one command. LEARN MORE → CONTENT OPTIMIZER Optimize existing blog posts for Google rankings and AI citations. Replace fabricated stats with sourced data. LEARN MORE → SEO CHECKER Validate on-page SEO elements: title tags, meta descriptions, heading hierarchy, internal links, and schema. LEARN MORE → FLOW FRAMEWORK 30 evidence-led blogging prompts across Find, Optimize, Win stages. Structured prompts, not improvised queries. CC BY 4.0. FLOW FRAMEWORK → MULTILINGUAL QA Audit a multilingual blog suite: completeness, hreflang correctness, locale-specific meta length, content freshness. MULTILINGUAL QA → VIEW ALL SKILLS → SCORE YOUR FIRST POST IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE SKOOL LINKEDIN SKILLS LLMS.TXT CHANGELOG LICENSE --- ## Content Optimizer - Rewrite Posts | Claude Blog URL: https://claude-blog.md/skills/blog-rewrite Content Optimizer - Rewrite Posts | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Content Optimizer // BLOG REWRITE REWRITE AND OPTIMIZE Use this skill in the hosted vault Improve an existing post for readers and search fundamentals. Claude Blog replaces unsupported claims with sourced facts, clarifies structure, checks crawlability, and aligns metadata and schema with visible content without promising rankings. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog rewrite COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS 5 OPTIMIZATION PASSES. When you run /blog rewrite, Claude Blog analyzes your existing post and applies five optimization passes. Each pass targets a different dimension of content quality, from factual accuracy to AI citation readiness. 01 STAT REPLACEMENT FINDS AND REPLACES FABRICATED OR OUTDATED STATISTICS WITH CURRENT, SOURCED DATA FROM TIER 1-3 PUBLICATIONS. 02 ANSWER-FIRST FORMAT RESTRUCTURES CONTENT WITH KEY TAKEAWAYS FIRST, THEN SUPPORTING DETAIL. OPTIMIZED FOR FEATURED SNIPPETS AND AI CITATIONS. 03 SCHEMA INJECTION ADDS ONLY PURPOSE-APPROPRIATE JSON-LD THAT MATCHES VISIBLE CONTENT. FAQPAGE IS OPTIONAL AND EARNS NO READINESS POINTS. 04 IMAGE SOURCING FINDS RELEVANT IMAGES FROM PIXABAY, UNSPLASH, AND PEXELS. GENERATES INLINE SVG CHARTS WHERE DATA EXISTS. 05 DUAL OPTIMIZATION EVERY REWRITE IS OPTIMIZED FOR GOOGLE RANKINGS (DECEMBER 2025 CORE UPDATE, E-E-A-T) AND AI CITATIONS (GEO/AEO). // USAGE HOW TO REWRITE A POST Step 1: Install Claude Blog git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git bash claude-blog/install.sh Step 2: Open Claude Code claude Step 3: Rewrite Your Post /blog rewrite Point it at any existing blog post. The optimizer runs all 5 passes and produces a reviewed version with supported claims, clearer structure, and purpose-appropriate metadata or schema. The score remains an internal editorial heuristic. // FAQ QUESTIONS ABOUT REWRITING What does the rewrite process change? - It can improve headings, replace unsupported claims with sourced facts, clarify the answer, add relevant media, and align metadata or schema with visible content. It does not insert FAQ markup or first-hand claims without a real need and evidence. Does it preserve my original voice? + Yes. The rewriter preserves the core message and voice while improving factual support, clarity, and technical consistency. Can it rewrite content from any CMS? + Yes. It works with HTML, Markdown, MDX, and plain text used by common CMS and static-site workflows. Does a rewrite guarantee a higher score or ranking? + No. The tool reports an internal editorial heuristic and specific findings. Any score change depends on the source material and accepted edits, and it does not predict Google rankings or AI citations. // RELATED SKILLS EXPLORE MORE CONTENT SCORING Score blog posts on a 100-point scale across 5 categories. Know exactly where your content stands before and after optimization. LEARN MORE → AI BLOG WRITER Write new SEO blog posts with a 4-agent pipeline. Research, draft, check, and review in one command. LEARN MORE → AI CITATION OPTIMIZER Optimize for Google AI Overviews, ChatGPT, Perplexity, and Bing Copilot. Passage-level citability scoring. LEARN MORE → FLOW FRAMEWORK 30 evidence-led blogging prompts across Find, Optimize, Win stages. Structured prompts, not improvised queries. CC BY 4.0. FLOW FRAMEWORK → GOOGLE APIS Search Console, GA4, PageSpeed, CrUX, and Indexing API in one skill. Real ranking and traffic data, free quota. GOOGLE APIS → VIEW ALL SKILLS → OPTIMIZE YOUR FIRST POST IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE SKOOL LINKEDIN SKILLS LLMS.TXT CHANGELOG LICENSE --- ## Blog Strategy Planner - AI Powered | Claude Blog URL: https://claude-blog.md/skills/blog-strategy Blog Strategy Planner - AI Powered | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Blog Strategy // BLOG STRATEGY AI BLOG STRATEGY Use this skill in the hosted vault Plan audience-led topic architecture, original-value opportunities, evidence requirements, content maintenance, distribution, search fundamentals, and honest AI-readiness review. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog strategy COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS STRATEGY FROM THE TERMINAL. When you run /blog strategy, Claude Blog builds a comprehensive content roadmap. It identifies topic clusters, maps your audience, analyzes competitors, and plans content that gets cited by AI search engines. 01 TOPIC CLUSTERS HUB-AND-SPOKE ARCHITECTURE. PILLAR PAGES LINKED TO SUPPORTING CONTENT. INTERNAL LINKING MAP GENERATED AUTOMATICALLY. 02 AUDIENCE MAPPING DEFINE TARGET PERSONAS, SEARCH INTENT CATEGORIES, AND CONTENT FUNNEL STAGES (AWARENESS, CONSIDERATION, DECISION). 03 COMPETITIVE ANALYSIS IDENTIFY CONTENT GAPS, RANKING OPPORTUNITIES, AND TOPICS YOUR COMPETITORS COVER THAT YOU DO NOT. 04 AI CITATION STRATEGY PLAN CONTENT THAT GETS CITED BY CHATGPT, PERPLEXITY, AND AI OVERVIEWS. PASSAGE-LEVEL CITABILITY PLANNING. 05 DISTRIBUTION CHANNELS YOUTUBE, REDDIT, REVIEW PLATFORMS, SOCIAL MEDIA. MAP CONTENT TO DISTRIBUTION CHANNELS FOR MAXIMUM REACH. // USAGE HOW TO PLAN YOUR STRATEGY Step 1: Install Claude Blog git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git bash claude-blog/install.sh Step 2: Open Claude Code claude Step 3: Build Your Strategy /blog strategy Provide your niche, target audience, and competitor URLs. The strategy planner builds topic clusters, maps audience personas, identifies content gaps, and produces a structured roadmap with distribution channels. // FAQ QUESTIONS ABOUT STRATEGY What does a blog strategy include? - A complete strategy covers audience needs, topic architecture, original-value opportunities, evidence requirements, a content calendar, maintenance, distribution, search fundamentals, and AI-readiness review. How does topic clustering work? + The strategy groups topics by shared intent and useful relationships, then plans contextual links between appropriate pages. It does not require a fixed number of supporting posts. Can it analyze my competitors? + Yes. Relevant competitor pages can reveal recurring coverage and gaps, but the goal is original reader value rather than copying formats or promising rankings. Does it plan for AI citations? + It plans for AI readiness through source fidelity, factual support, clear entities, crawlability, and useful original material. It does not predict or guarantee citations. // RELATED SKILLS EXPLORE MORE CONTENT CALENDAR Generate editorial calendars from your strategy. Weekly and monthly publishing schedules with topic assignments. LEARN MORE → CONTENT BRIEF Create detailed content briefs for each post in your strategy. Target keywords, outline, sources, and word count. LEARN MORE → BLOG OUTLINE Generate detailed post outlines with heading hierarchy, key points, and source suggestions for each section. LEARN MORE → FLOW FRAMEWORK 30 evidence-led blogging prompts across Find, Optimize, Win stages. Structured prompts, not improvised queries. CC BY 4.0. FLOW FRAMEWORK → TOPIC CLUSTERS SERP-overlap clustering, not text similarity. Hub-and-spoke architecture for topical authority. Community-built. TOPIC CLUSTERS → NOTEBOOKLM RESEARCH Source-grounded research from a curated corpus. Citations stay locked to your trusted documents, no model hallucination. NOTEBOOKLM RESEARCH → VIEW ALL SKILLS → PLAN YOUR STRATEGY IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE SKOOL LINKEDIN SKILLS LLMS.TXT CHANGELOG LICENSE --- ## AI Blog Image Generator - Gemini | Claude Blog URL: https://claude-blog.md/skills/blog-image AI Blog Image Generator - Gemini | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/AI Image Generation // BLOG IMAGE AI IMAGE GENERATION Use this skill in the hosted vault Generate blog hero images, OG graphics, and inline illustrations from text prompts. Powered by Google Gemini via the Banana MCP server. Dark aesthetic with coral accents by default. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog image COPY REQUIRES CLAUDE BLOG + BANANA MCP INSTALLED // HOW IT WORKS GEMINI-POWERED IMAGES. When you run /blog image, Claude Blog generates blog-optimized images using Google Gemini. It uses a 5-component prompt formula optimized for tech and marketing content, then automatically converts output to WebP for web performance. 01 GEMINI POWERED GENERATES IMAGES VIA GOOGLE GEMINI 3.1 FLASH IMAGE PREVIEW. HIGH QUALITY, FAST, AND FREE WITH A GOOGLE API KEY. 02 BLOG-OPTIMIZED HERO IMAGES IN 16:9 (1920x1080), OG IMAGES IN 1200x630. AUTOMATIC WEBP CONVERSION AND COMPRESSION. 03 PROMPT ENGINEERING 5-COMPONENT PROMPT FORMULA: SUBJECT, ACTION, CONTEXT, COMPOSITION, STYLE. OPTIMIZED FOR TECH AND MARKETING CONTENT. 04 BATCH GENERATION GENERATE HERO AND OG IMAGES FOR MULTIPLE POSTS IN A SINGLE SESSION. CONSISTENT VISUAL STYLE ACROSS YOUR BLOG. // USAGE HOW TO GENERATE IMAGES Step 1: Install Claude Blog git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git bash claude-blog/install.sh Step 2: Set Up Banana MCP (for Gemini) Blog image generation requires the Banana MCP server for Gemini API access. Follow the Banana Claude setup guide and add your Google API key. Step 3: Generate Images /blog image Describe your image or let Claude Blog generate a prompt from your post content. Images are saved as WebP in your blog assets directory. // FAQ QUESTIONS ABOUT IMAGES What image formats does it generate? - It generates PNG images that are automatically converted to WebP for web use. Hero images are 16:9 ratio (1920x1080) and OG images are 1200x630 for social sharing. Is the Gemini API free? + Google offers a free tier for the Gemini API with generous limits. You need a Google API key, which is free to create. No credit card required for basic usage. Can I customize the visual style? + Yes. You can specify custom prompts, aspect ratios, and style directions. The default style is dark background with coral accent, minimal tech aesthetic, but you can override any of these. Does it require the Banana extension? + Yes. Blog image uses the Banana MCP server (nanobanana-mcp) for Gemini integration. Install it separately or use the direct API script as a fallback. // RELATED SKILLS EXPLORE MORE AI BLOG WRITER Write SEO blog posts with a 4-agent pipeline. Research, draft, check, and review in one command. LEARN MORE → CONTENT REPURPOSE Repurpose blog posts into social media threads, newsletters, video scripts, and slide decks. LEARN MORE → SEO CHECKER Validate on-page SEO elements including image alt text, OG tags, and structured data for images. LEARN MORE → GEMINI TTS AUDIO 30-voice text-to-speech in 80+ languages. Generate podcast versions of blog posts at 24 kHz studio quality. GEMINI TTS AUDIO → VIEW ALL SKILLS → GENERATE YOUR FIRST IMAGE IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE SKOOL LINKEDIN SKILLS LLMS.TXT CHANGELOG LICENSE --- ## AI Readiness Heuristic - GEO Audit | Claude Blog URL: https://claude-blog.md/skills/blog-geo AI Readiness Heuristic - GEO Audit | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/AI Citation Optimizer // BLOG GEO AI CITATION READINESS Use this skill in the hosted vault Review source fidelity, factual support, entity clarity, crawlability, visible-content consistency, and reader usefulness. The 0-100 result is an internal readiness heuristic, not a citation probability or guarantee. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog geo COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS ONE HONEST READINESS AUDIT. When you run /blog geo, Claude Blog applies an internal editorial heuristic. It reviews sources, supported claims, entity clarity, crawlability, and visible-content consistency without pretending to predict platform selection. 01 READINESS HEURISTIC A 0-100 INTERNAL EDITORIAL SCORE. IT IS NOT A CALIBRATED PROBABILITY, RANKING FACTOR, OR PLATFORM PROMISE. 02 CITABILITY ANALYSIS PASSAGE-LEVEL SCORING. IDENTIFIES WHICH PARAGRAPHS ARE QUOTABLE AND WHICH NEED RESTRUCTURING FOR AI EXTRACTION. 03 STRUCTURAL READABILITY CHECKS HEADING HIERARCHY, LIST USAGE, TABLE STRUCTURE, AND DEFINITION FORMATTING FOR AI PARSING. 04 ENTITY CLARITY ENSURES BRAND NAMES, PRODUCT NAMES, AND TECHNICAL TERMS ARE UNAMBIGUOUS AND CONSISTENTLY REFERENCED. 05 CRAWLER ACCESS REVIEWS APPLICABLE ROBOTS DIRECTIVES, RENDERED ACCESS, AND TECHNICAL DISCOVERY CONTEXT WITHOUT PROMISING INCLUSION. // USAGE HOW TO OPTIMIZE FOR AI Step 1: Install Claude Blog git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git bash claude-blog/install.sh Step 2: Open Claude Code claude Step 3: Run the GEO Audit /blog geo Point it at any blog post. You get an internal readiness score, source and claim checks, entity clarity findings, and practical recommendations. No fixed passage length, FAQ, or schema type earns points by itself. // FAQ QUESTIONS ABOUT GEO What is the AI-readiness heuristic? - It is an internal editorial review of source fidelity, factual support, entity clarity, crawlability, visible-content consistency, and reader usefulness. It does not predict or guarantee citations. Does it calculate a citation probability? + No. The 0-100 output is a readiness heuristic. Platform behavior changes and no passage length, heading pattern, FAQ, or schema type can guarantee selection by an AI system. How do I improve readiness? + Publish original, useful material; support factual claims with suitable sources; make important entities and relationships clear; keep key content crawlable; and ensure metadata and structured data match visible content. Does it check crawler access? + Yes. It reviews applicable robots directives and technical accessibility. Crawler access and llms.txt are reported as technical context, not as guarantees of inclusion or citation. // RELATED SKILLS EXPLORE MORE CONTENT SCORING Score blog posts on a 100-point scale across 5 categories including AI citation readiness as a weighted component. LEARN MORE → CONTENT OPTIMIZER Improve existing posts for usefulness, source fidelity, search fundamentals, and AI readiness in one rewrite pass. LEARN MORE → SEO CHECKER Validate on-page SEO elements: title tags, meta descriptions, heading hierarchy, schema, and crawler directives. LEARN MORE → FLOW FRAMEWORK 30 evidence-led blogging prompts across Find, Optimize, Win stages. Structured prompts, not improvised queries. CC BY 4.0. FLOW FRAMEWORK → TOPIC CLUSTERS SERP-overlap clustering, not text similarity. Hub-and-spoke architecture for topical authority. Community-built. TOPIC CLUSTERS → VIEW ALL SKILLS → REVIEW AI READINESS IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE SKOOL LINKEDIN SKILLS LLMS.TXT CHANGELOG LICENSE --- ## Blog Health Audit - Full Site | Claude Blog URL: https://claude-blog.md/skills/blog-audit Blog Health Audit - Full Site | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Blog Audit // SKILL FULL-SITE BLOG AUDIT Use this skill in the hosted vault Scan your entire blog directory. Get per-post quality scores, find orphan pages, detect topic cannibalization, and identify stale content. The audit produces a prioritized action queue so you know exactly what to fix first. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog audit ./content/blog COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS SCAN. SCORE. PRIORITIZE. When you run /blog audit, Claude Blog scans every post in your blog directory. Each post is evaluated for frontmatter, usefulness, source and claim quality, links, heading logic, metadata, intent-dependent completeness, readability context, media, and technical consistency. Every post receives an individual score from 0 to 100. The audit then aggregates findings into a site-wide health report with orphan page detection, cannibalization warnings, and stale content flags. You get a prioritized action queue sorted by impact. 01 PER-POST SCORING Every post gets an internal 0-100 editorial score based on usefulness, evidence, headings, links, metadata, readability context, media, and technical consistency. Word count alone earns or loses no points. 02 ORPHAN DETECTION Finds blog posts with zero internal links pointing to them. Orphan pages are invisible to crawlers and readers. The audit flags every orphan and suggests linking opportunities. 03 CANNIBALIZATION Detects posts competing for the same keywords and search intent. Clusters semantically similar titles and headings, then recommends whether to merge, redirect, or differentiate. 04 STALE CONTENT Identifies posts that have not been updated in over 12 months, contain outdated statistics, or reference deprecated tools. Flags content decay before it hurts your rankings. // USAGE HOW TO RUN AN AUDIT Step 1: Install Claude Blog git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh Step 2: Open Claude Code claude Step 3: Run the Audit /blog audit ./content/blog Point it at your blog directory. The audit scans every markdown or HTML file, scores each post, and outputs a prioritized report. // FAQ QUESTIONS ABOUT THE AUDIT What does the audit scan? - It checks frontmatter, usefulness, source and claim quality, links, heading logic, metadata, intent-dependent completeness, readability context, media, and technical consistency. Each post receives an internal 0-100 editorial score. How many posts can it audit? + There is no hard limit. It processes posts sequentially, so runtime depends on post count, content size, and enabled checks. What is an orphan page? + An orphan page has no internal links pointing to it from other pages. The audit flags these pages so you can assess whether a contextual link, consolidation, or removal is appropriate. How does it detect cannibalization? + The audit compares topics and search intent across posts, then flags meaningful overlap for human review. It suggests whether to differentiate, consolidate, redirect, or leave the pages separate. // RELATED SKILLS EXPLORE MORE BLOG ANALYZE Deep analysis of individual posts with readability context, evidence checks, and content-quality assessment. LEARN MORE → KEYWORD CANNIBALIZATION Dedicated cannibalization detector that clusters semantically similar targets and flags competing posts. LEARN MORE → BLOG STRATEGY Content strategy planning with topic clusters, competitive gaps, and publishing roadmaps. LEARN MORE → FLOW FRAMEWORK 30 evidence-led blogging prompts across Find, Optimize, Win stages. Structured prompts, not improvised queries. CC BY 4.0. FLOW FRAMEWORK → GOOGLE APIS Search Console, GA4, PageSpeed, CrUX, and Indexing API in one skill. Real ranking and traffic data, free quota. GOOGLE APIS → MULTILINGUAL QA Audit a multilingual blog suite: completeness, hreflang correctness, locale-specific meta length, content freshness. MULTILINGUAL QA → VIEW ALL SKILLS → AUDIT YOUR BLOG IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Content Brief Generator - AI | Claude Blog URL: https://claude-blog.md/skills/blog-brief Content Brief Generator - AI | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Content Briefs // SKILL AI CONTENT BRIEFS Use this skill in the hosted vault Generate detailed briefs before writing. Audience intent, competitive gaps, evidence needs, relevant media, intent-dependent coverage, contextual internal links, and template recommendations. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog brief "how to optimize blog SEO" COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS BRIEF BEFORE YOU WRITE. Running /blog brief with a target topic generates a structured plan. It reviews relevant competing content, identifies opportunities for original value, and maps headings, evidence needs, useful media, and intent-dependent coverage. Every brief includes internal linking targets from your existing blog content, so new posts strengthen your site architecture from day one. 01 KEYWORD TARGETING Primary and secondary keyword recommendations based on topic analysis. Includes search intent classification (informational, transactional, navigational) and keyword difficulty estimates. 02 COMPETITIVE GAPS Analyzes top-ranking content for your target topic. Identifies subtopics, questions, and angles that competitors miss, giving you clear differentiation opportunities. 03 STAT RECOMMENDATIONS Suggests specific statistics, data points, and research findings to include in your post. Data-backed content earns more backlinks and higher trust signals. 04 TEMPLATE SELECTION Recommends the best post template for your topic: how-to guide, listicle, comparison, case study, or pillar page. Each template includes section-by-section structure. // USAGE HOW TO GENERATE A BRIEF Step 1: Install Claude Blog git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh Step 2: Open Claude Code claude Step 3: Generate a Brief /blog brief "your target topic or keyword" The brief generator produces a markdown file with everything you need to start writing: keywords, outline, competitive analysis, and content recommendations. // FAQ QUESTIONS ABOUT BRIEFS What goes into a content brief? - A brief includes the audience task, primary topic and related concepts, an intent-led outline, useful competitive gaps, claims that need evidence, relevant media, contextual internal links, and a suitable template. How does competitive analysis work? + The generator reviews relevant search results for recurring topics, evidence, and media, then identifies opportunities for original value. It does not copy competitors or promise rankings. Can I customize the brief? + Yes. Specify the audience, topic, content type, tone, constraints, and available evidence. Coverage adapts to intent rather than a fixed word-count range. Does it suggest images? + Yes. It recommends relevant screenshots, diagrams, charts, or illustrations and appropriate alternative text. Media should help the reader, not satisfy a quota. // RELATED SKILLS EXPLORE MORE BLOG WRITE AI-powered blog post writing with SEO optimization, persona-aware tone, and automatic internal linking. LEARN MORE → CONTENT OUTLINES SERP-informed outlines with logical headings, intent-dependent coverage, and optional question research. LEARN MORE → BLOG STRATEGY Content strategy planning with topic clusters, competitive gaps, and publishing roadmaps. LEARN MORE → NOTEBOOKLM RESEARCH Source-grounded research from a curated corpus. Citations stay locked to your trusted documents, no model hallucination. NOTEBOOKLM RESEARCH → FLOW FRAMEWORK 30 evidence-led blogging prompts across Find, Optimize, Win stages. Structured prompts, not improvised queries. CC BY 4.0. FLOW FRAMEWORK → VIEW ALL SKILLS → GENERATE YOUR FIRST BRIEF IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Editorial Calendar - AI Planner | Claude Blog URL: https://claude-blog.md/skills/blog-calendar Editorial Calendar - AI Planner | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Editorial Calendar // SKILL EDITORIAL CALENDAR Use this skill in the hosted vault Plan your publishing schedule from the terminal. Monthly and quarterly calendars with topic clusters, content mix formulas, seasonal opportunities, and distribution scheduling. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog calendar --quarter Q2 --posts-per-week 3 COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS PLAN. CLUSTER. PUBLISH. Running /blog calendar generates a structured publishing plan based on your niche, existing content, and publishing cadence. Topics are organized into clusters so each post strengthens the others through internal linking. The calendar balances content types and highlights relevant seasonal opportunities. Every entry includes a title idea, audience need, topic, content type, evidence requirements, and estimated effort. 01 TOPIC CLUSTERS Groups related topics into clusters with pillar pages and supporting posts. Each cluster builds topical authority and creates natural internal linking pathways. 02 CONTENT MIX Balances content types across your calendar: how-to guides, listicles, case studies, comparisons, and opinion pieces. Customizable ratios ensure variety without sacrificing focus. 03 SEASONAL PLANNING Identifies seasonal trends and industry events relevant to your niche. Schedules timely content 4-6 weeks before peak interest so you rank when traffic surges. 04 DISTRIBUTION SCHEDULE Pairs each blog post with distribution recommendations: social media platforms, email newsletter timing, community posting, and syndication opportunities. // USAGE HOW TO PLAN YOUR CALENDAR Step 1: Install Claude Blog git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh Step 2: Open Claude Code claude Step 3: Generate a Calendar /blog calendar --quarter Q2 --posts-per-week 3 Specify your planning horizon and publishing frequency. The calendar adapts to your niche based on existing blog content and generates a complete publishing roadmap. // FAQ QUESTIONS ABOUT CALENDARS What does a calendar include? - An editorial calendar includes dates, audience needs, topic clusters, content types, evidence requirements, contextual internal links, maintenance work, and distribution ideas. Estimated effort and coverage adapt to each topic. How far ahead can it plan? + It supports monthly, quarterly, and annual horizons. Plans should be revisited when priorities, evidence, seasonality, or search behavior changes. Does it detect content decay? + It can use performance and substantive content-change evidence to schedule refresh reviews. A metadata date alone does not prove freshness or decay. Can I customize the mix? + Yes. Set a content mix that fits the audience and business purpose. The calendar preserves topical coherence without requiring raw word-count targets. // RELATED SKILLS EXPLORE MORE BLOG STRATEGY Content strategy planning with topic clusters, competitive gaps, and publishing roadmaps. LEARN MORE → CONTENT BRIEFS Generate detailed briefs with target keywords, competitive analysis, and content outlines. LEARN MORE → CONTENT OUTLINES SERP-informed content outlines with logical headings, intent-dependent coverage, and optional question research. LEARN MORE → TOPIC CLUSTERS SERP-overlap clustering, not text similarity. Hub-and-spoke architecture for topical authority. Community-built. TOPIC CLUSTERS → VIEW ALL SKILLS → PLAN YOUR CONTENT IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Blog Schema Generator - JSON-LD | Claude Blog URL: https://claude-blog.md/skills/blog-schema Blog Schema Generator - JSON-LD | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Schema Generator // SKILL BLOG SCHEMA GENERATOR Use this skill in the hosted vault Generate complete JSON-LD structured data for every blog post. BlogPosting, FAQPage, Person, Organization, BreadcrumbList, and ImageObject. Validates against current Google requirements and warns about deprecated types. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog schema ./content/blog/my-post.md COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS STRUCTURED DATA. AUTOMATED. Running /blog schema on a blog post extracts metadata from frontmatter, headings, and content body. It generates valid JSON-LD blocks for every applicable schema type and validates each one against Google's current requirements. The generator warns about deprecated schema types and restricted rich result eligibility so you never waste effort on markup that will not produce results. 01 BLOGPOSTING SCHEMA Complete BlogPosting markup with headline, datePublished, dateModified, author, publisher, image, description, wordCount, and articleBody excerpt. All required and recommended fields. 02 FAQPAGE SCHEMA Generates optional FAQPage markup only for visible FAQs. It earns no Google or AI-readiness points and is never inserted as a quota. 03 PERSON SCHEMA Author schema with name, url, sameAs links (social profiles), jobTitle, and knowsAbout topics. Builds author entity signals that strengthen E-E-A-T across all posts. 04 VALIDATION Validates every generated block against Google's structured data requirements. Checks required fields, data types, URL formats, and date formats. Reports errors, warnings, and recommendations. // USAGE HOW TO GENERATE SCHEMA Step 1: Install Claude Blog git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh Step 2: Open Claude Code claude Step 3: Generate Schema /blog schema ./content/blog/my-post.md Point it at any blog post. The generator reads frontmatter and content, produces JSON-LD blocks, validates them, and outputs ready-to-paste markup. // FAQ QUESTIONS ABOUT SCHEMA Which schema types does it generate? - The generator can produce BlogPosting, optional FAQPage for visible FAQs, Person, Organization, BreadcrumbList, and ImageObject. It recommends only types that describe the page's visible content and purpose. Does it validate against Google? + Yes. It checks current Google requirements, visible-content consistency, and retired or restricted rich-result types. Structured data is an eligibility aid, not a ranking guarantee. Can it add rendered JSON-LD? + Yes. Server-rendered or JavaScript-generated JSON-LD is acceptable when it reaches the rendered DOM, matches visible content, and passes validation. Does FAQPage earn Google or AI-readiness points? + No. FAQPage is optional markup for a visible FAQ. It provides no Google rich-result or AI-readiness points here. QAPage is reserved for genuine single-question community-answer pages. // RELATED SKILLS EXPLORE MORE SEO CHECKER Post-writing SEO validation with pass/fail checklist covering title tags, meta descriptions, and OG tags. LEARN MORE → BLOG ANALYZE Deep analysis of individual posts with readability context, evidence checks, and content-quality assessment. LEARN MORE → BLOG WRITE AI-powered blog post writing with SEO optimization, persona-aware tone, and automatic internal linking. LEARN MORE → GOOGLE APIS Search Console, GA4, PageSpeed, CrUX, and Indexing API in one skill. Real ranking and traffic data, free quota. GOOGLE APIS → MULTILINGUAL QA Audit a multilingual blog suite: completeness, hreflang correctness, locale-specific meta length, content freshness. MULTILINGUAL QA → VIEW ALL SKILLS → GENERATE SCHEMA IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Content Repurposing - Multi-Platform | Claude Blog URL: https://claude-blog.md/skills/blog-repurpose Content Repurposing - Multi-Platform | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Content Repurposing // SKILL CONTENT REPURPOSING Use this skill in the hosted vault One blog post, five platforms. Transform any article into Twitter/X threads, LinkedIn articles, YouTube video scripts, Reddit discussion posts, and email newsletter excerpts. Each format adapts tone and structure for its platform. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog repurpose ./content/blog/my-post.md --all COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS ONE POST. FIVE PLATFORMS. Running /blog repurpose reads your blog post, extracts the core message and key points, then generates platform-specific content for each channel. Every output adapts tone, length, and structure to match platform conventions. The repurposer preserves your original message and data points while reformatting for maximum engagement on each platform. No content is fabricated. 01 TWITTER THREADS 5-15 tweet threads with a compelling hook, numbered insights, and a call-to-action linking back to the full post. Optimized for engagement and retweets. 02 LINKEDIN ARTICLES Professional long-form articles with engagement-optimized formatting: short paragraphs, bold key points, and strategic line breaks. 800-1200 words. 03 YOUTUBE SCRIPTS Video scripts with intro hook, section-by-section talking points, B-roll suggestions, and closing CTA. Timed for 5-10 minutes of speaking. 04 REDDIT POSTS Discussion-style posts adapted for subreddit norms. Conversational tone, value-first approach, and community-appropriate formatting. 300-600 words. 05 EMAIL NEWSLETTERS Newsletter excerpts with subject line, preview text, key takeaways, and a clear link to the full post. Direct tone optimized for click-through. // USAGE HOW TO REPURPOSE CONTENT Step 1: Install Claude Blog git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh Step 2: Open Claude Code claude Step 3: Repurpose a Post /blog repurpose ./content/blog/my-post.md --all Use --all for all platforms, or specify individual ones: --twitter, --linkedin, --youtube, --reddit, --email. // FAQ QUESTIONS ABOUT REPURPOSING Which platforms does it support? - The repurposing skill supports five platforms: Twitter/X (thread format with hooks and numbered tweets), LinkedIn (professional article with engagement-optimized formatting), YouTube (video script with intro, sections, and CTA), Reddit (discussion-style post adapted for subreddit norms), and Email (newsletter excerpt with subject line and preview text). Does it preserve the original message? + Yes. The repurposer extracts the core message, key points, and supporting evidence from your blog post. It restructures and adapts the tone for each platform while keeping the central argument and data points intact. Nothing is fabricated. Can I customize the tone per platform? + Yes. Each platform output uses a default tone profile (casual for Twitter, professional for LinkedIn, conversational for YouTube, community-native for Reddit, direct for email), but you can override these with your own writing persona or tone preferences. How long are the outputs? + Outputs are sized for each platform: Twitter threads are 5-15 tweets, LinkedIn articles are 800-1200 words, YouTube scripts are 5-10 minutes of speaking time, Reddit posts are 300-600 words, and email excerpts are 200-400 words with a link back to the full post. // RELATED SKILLS EXPLORE MORE BLOG WRITE AI-powered blog post writing with SEO optimization, persona-aware tone, and automatic internal linking. LEARN MORE → BLOG STRATEGY Content strategy planning with topic clusters, competitive gaps, and publishing roadmaps. LEARN MORE → EDITORIAL CALENDAR Generate publishing schedules with topic clusters, content mix, and seasonal opportunities. LEARN MORE → GEMINI TTS AUDIO 30-voice text-to-speech in 80+ languages. Generate podcast versions of blog posts at 24 kHz studio quality. GEMINI TTS AUDIO → MULTILINGUAL PUBLISHING One command, every language. Source post plus translated and localized variants plus hreflang in a single pipeline. MULTILINGUAL PUBLISHING → VIEW ALL SKILLS → REPURPOSE YOUR CONTENT IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Content Outline - SERP Informed | Claude Blog URL: https://claude-blog.md/skills/blog-outline Content Outline - SERP Informed | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Content Outlines // SKILL SERP-INFORMED OUTLINES Use this skill in the hosted vault Generate article outlines informed by relevant search results. Logical headings, intent-dependent coverage, evidence and media markers, optional question research, and contextual internal links. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog outline "best blog SEO tools 2026" COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS OUTLINES THAT RANK. Running /blog outline with a target keyword generates a structured outline informed by what currently ranks in search results. The outline identifies heading patterns, content depth, and question opportunities from top-performing content. Every section includes a purpose, evidence needs, media opportunities, and contextual internal-link suggestions. Coverage expands only as the reader's intent and subject require. 01 HEADING HIERARCHY Complete H2/H3 heading structure with descriptive section summaries. Each heading is informed by competitive analysis and search intent patterns. 02 INTENT-LED COVERAGE Section depth follows the question, evidence, and reader task. Short complete answers are not padded to imitate competing pages. 03 OPTIONAL QUESTION RESEARCH Relevant public questions can inform coverage when they help the reader. No FAQ section or question-heading quota is required. 04 COMPETITIVE GAPS Identifies subtopics and angles that top-ranking pages miss. These gaps become your differentiation points and are highlighted in the outline structure. // USAGE HOW TO GENERATE AN OUTLINE Step 1: Install Claude Blog git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh Step 2: Open Claude Code claude Step 3: Generate an Outline /blog outline "your target keyword" The generator produces a structured markdown file with headings, section purposes, evidence notes, optional questions, and internal-link opportunities. Edit it before passing it to the writing skill. // FAQ QUESTIONS ABOUT OUTLINES How does SERP analysis work? - The outline generator reviews the structure and topics of relevant search results. It identifies recurring coverage and useful gaps, then treats those findings as research rather than a formula to copy. What does the outline include? + Each outline includes a title recommendation, logical heading hierarchy, section purpose and evidence notes, relevant media markers, optional question research, and contextual internal-link opportunities. Coverage adapts to intent instead of a fixed word target. Can I edit the outline before writing? + Yes. The outline is a structured markdown file that you can edit before writing. Add, remove, or reorder sections to fit the reader's needs. Does it plan internal links? + Yes. It identifies contextually relevant existing posts and useful locations for internal links. There is no arbitrary link quota. // RELATED SKILLS EXPLORE MORE CONTENT BRIEFS Generate detailed briefs with target keywords, competitive analysis, and content outlines. LEARN MORE → BLOG WRITE AI-powered blog post writing with SEO optimization, persona-aware tone, and automatic internal linking. LEARN MORE → BLOG STRATEGY Content strategy planning with topic clusters, competitive gaps, and publishing roadmaps. LEARN MORE → NOTEBOOKLM RESEARCH Source-grounded research from a curated corpus. Citations stay locked to your trusted documents, no model hallucination. NOTEBOOKLM RESEARCH → TOPIC CLUSTERS SERP-overlap clustering, not text similarity. Hub-and-spoke architecture for topical authority. Community-built. TOPIC CLUSTERS → FLOW FRAMEWORK 30 evidence-led blogging prompts across Find, Optimize, Win stages. Structured prompts, not improvised queries. CC BY 4.0. FLOW FRAMEWORK → VIEW ALL SKILLS → BUILD YOUR OUTLINE IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Blog SEO Checker - Post Validation | Claude Blog URL: https://claude-blog.md/skills/blog-seo-check Blog SEO Checker - Post Validation | Claude Blog Skip to main content CLAUDE BLOG HOMESKILLSBLOG HOMESKILLSBLOGGITHUB Home/Skills/SEO Checker // SKILL POST-WRITING SEO CHECK Use this skill in the hosted vault Run a post-writing SEO checklist for descriptive titles and metadata, logical headings, contextual links, canonical URLs, schema consistency, social metadata, and crawlability. It uses no keyword-density or minimum-link quota. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog seo-check ./content/blog/my-post.md COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS PASS/FAIL. NO GUESSING. Running /blog seo-check on a blog post produces a structured pass/fail checklist. Every on-page SEO element is validated against current best practices. Green checks for passing items, red flags for failures, with specific fix recommendations for each issue. The checker covers everything from title tag length to image alt text completeness, so no SEO detail is missed before publishing. 01 TITLE AND META Validates title tag length (under 60 characters), meta description quality (120-160 characters), keyword placement in title, and unique title across your blog. 02 HEADING HIERARCHY Checks for single H1, logical H2/H3 nesting without skipped levels, keyword presence in headings, and heading length optimization for featured snippets. 03 LINK AUDIT Reviews whether internal and external links are useful in context, checks for broken links, and verifies safe external-link attributes. There is no arbitrary minimum. 04 OG AND SCHEMA Validates Open Graph tags (og:title, og:description, og:image), Twitter Card markup, canonical URL, and JSON-LD schema presence and correctness. // USAGE HOW TO CHECK YOUR POST Step 1: Install Claude Blog git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh Step 2: Open Claude Code claude Step 3: Run the SEO Check /blog seo-check ./content/blog/my-post.md Run this after writing, before publishing. The checker produces a pass/fail report with specific fix recommendations for every failing item. // FAQ QUESTIONS ABOUT THE CHECKER What does the SEO check cover? - It checks whether titles and descriptions are descriptive and useful, headings are logical, links are contextual, canonical and social metadata are present, structured data matches visible content, URLs are clear, images have appropriate alternatives, and important content is crawlable. Length ranges are preview guidance, not ranking rules. There is no keyword-density or minimum-link quota. Is it different from the full audit? + Yes. The full blog audit scans an entire directory and reports site-wide concerns such as orphan detection and cannibalization. The SEO checker focuses on one post and its on-page and technical signals. Can it auto-fix issues? + The checker identifies issues and provides recommendations, but it does not modify files automatically. You review each finding and decide whether it is appropriate for the page's purpose. Does it check mobile compatibility? + It validates viewport metadata, responsive image attributes, and basic accessibility. Full rendering and interaction tests require a browser-based audit. // RELATED SKILLS EXPLORE MORE BLOG ANALYZE Deep analysis of individual posts with readability context, evidence checks, and content-quality assessment. LEARN MORE → SCHEMA GENERATOR Generate JSON-LD schema markup for blog posts. BlogPosting, FAQPage, Person, and BreadcrumbList. LEARN MORE → BLOG WRITE AI-powered blog post writing with SEO optimization, persona-aware tone, and automatic internal linking. LEARN MORE → FLOW FRAMEWORK 30 evidence-led blogging prompts across Find, Optimize, Win stages. Structured prompts, not improvised queries. CC BY 4.0. FLOW FRAMEWORK → GOOGLE APIS Search Console, GA4, PageSpeed, CrUX, and Indexing API in one skill. Real ranking and traffic data, free quota. GOOGLE APIS → MULTILINGUAL QA Audit a multilingual blog suite: completeness, hreflang correctness, locale-specific meta length, content freshness. MULTILINGUAL QA → VIEW ALL SKILLS → CHECK YOUR POST IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Writing Personas - Voice Profiles | Claude Blog URL: https://claude-blog.md/skills/blog-persona Writing Personas - Voice Profiles | Claude Blog Skip to main content CLAUDE BLOG HOMESKILLSBLOG HOMESKILLSBLOGGITHUB Home/Skills/Writing Personas // SKILL WRITING PERSONAS Use this skill in the hosted vault Create custom writing profiles so every post sounds like you. The NNGroup 4-dimension tone framework defines your voice: Funny-Serious, Formal-Casual, Respectful-Irreverent, Enthusiastic-Matter-of-fact. Set readability targets, vocabulary tier, and contraction frequency. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog persona create "Technical Writer" COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS YOUR VOICE. EVERY POST. Running /blog persona create walks you through defining a writing voice profile. You set tone dimensions, readability targets, vocabulary preferences, and sentence structure patterns. The persona is saved as a JSON file and can be applied to any writing or rewriting task. When a persona is active, all content generation and analysis tools adapt their behavior to match your defined voice. Consistency across dozens or hundreds of posts becomes automatic. 01 4-DIMENSION TONE Define your voice across four NNGroup axes: Funny-Serious, Formal-Casual, Respectful-Irreverent, and Enthusiastic-Matter-of-fact. Each scored 1-5 for precise control. 02 READABILITY TARGETS Set your target Flesch-Kincaid grade level, average sentence length, and paragraph length. The writing skill enforces these targets automatically. 03 VOCABULARY CONTROL Choose vocabulary tier (basic, intermediate, advanced, technical) and set contraction frequency, jargon tolerance, and banned word lists for brand consistency. 04 PERSONA LIBRARY Store multiple personas for different content types or audiences. Switch between "Technical Writer," "Marketing Lead," or "Casual Blogger" with a single parameter. // USAGE HOW TO CREATE A PERSONA Step 1: Install Claude Blog git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh Step 2: Open Claude Code claude Step 3: Create a Persona /blog persona create "Technical Writer" Follow the prompts to set tone dimensions, readability targets, and vocabulary preferences. The persona is saved and ready to use with any writing or analysis command. // FAQ QUESTIONS ABOUT PERSONAS What is the NNGroup tone framework? - The Nielsen Norman Group 4-dimension tone framework defines writing voice across four axes: Funny vs Serious, Formal vs Casual, Respectful vs Irreverent, and Enthusiastic vs Matter-of-fact. Each axis is scored from 1 to 5. A persona might be Serious (2), Casual (4), Respectful (4), Enthusiastic (3), creating a friendly but professional voice. How many personas can I create? + There is no limit. Create as many personas as you need for different content types, audiences, or brand voices. Personas are stored as JSON files in your project directory and can be shared across team members or projects. Can I apply personas to rewrites? + Yes. The rewrite skill accepts a persona parameter that transforms existing content to match the specified voice profile. This is useful for adapting guest posts, refreshing old content, or aligning acquired content with your brand voice. Does it affect scoring? + Yes. When a persona is active, the blog analyzer scores content against the persona's readability target, vocabulary tier, and sentence length distribution. Posts that deviate from the persona's guidelines receive lower consistency scores and specific recommendations to align with the target voice. // RELATED SKILLS EXPLORE MORE BLOG WRITE AI-powered blog post writing with SEO optimization, persona-aware tone, and automatic internal linking. LEARN MORE → BLOG REWRITE Rewrite and refresh existing blog posts with updated information, improved structure, and persona alignment. LEARN MORE → BLOG ANALYZE Deep analysis of individual posts with readability context, evidence checks, and content-quality assessment. LEARN MORE → SEO TRANSLATION Format-preserving translation: markdown, frontmatter, schema, locale-correct number and date formatting intact. SEO TRANSLATION → CULTURAL LOCALIZE Beyond translation: idiom mapping, example substitution, legal-reference swaps for DACH, Francophone, Hispanic, Japanese. CULTURAL LOCALIZE → VIEW ALL SKILLS → CREATE YOUR VOICE IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Fact Checker - Source Verification | Claude Blog URL: https://claude-blog.md/skills/blog-factcheck Fact Checker - Source Verification | Claude Blog Skip to main content CLAUDE BLOG HOMESKILLSBLOG HOMESKILLSBLOGGITHUB Home/Skills/Fact Checker // SKILL FACT CHECKING Use this skill in the hosted vault Verify every statistic and claim in your blog post. The fact checker extracts all statistical claims, fetches each cited source URL, and scores match confidence. Exact matches score 1.0, approximate matches lower. No more fabricated data. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog factcheck ./content/blog/my-post.md COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS VERIFY. SCORE. TRUST. Running /blog factcheck scans your blog post for every statistical claim, percentage, date, and factual assertion. For each claim, it identifies the cited source, fetches the page, and compares the data. You get a confidence score for every claim in your post. Claims without sources are flagged for manual addition. Broken source URLs trigger alternative source suggestions. The result is a blog post where every number is verifiable. 01 CLAIM EXTRACTION Scans your post for all statistical claims: numbers, percentages, dates, rankings, and factual assertions. Each claim is isolated and paired with its cited source URL. 02 SOURCE FETCHING Fetches each cited URL, parses the content, and searches for the specific data point. Handles HTML, PDF, and plain text sources. Reports inaccessible URLs. 03 CONFIDENCE SCORING Each claim receives a 0.0 to 1.0 confidence score. 1.0 means exact match found in source. Below 0.5 means the claim needs manual verification or a better source. 04 REPORT GENERATION Produces a verification report listing every claim, its source, confidence score, and recommendation. Unsourced claims and broken URLs get alternative source suggestions. // USAGE HOW TO FACT CHECK Step 1: Install Claude Blog git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh Step 2: Open Claude Code claude Step 3: Run the Fact Checker /blog factcheck ./content/blog/my-post.md The fact checker scans your post, fetches all cited sources, and produces a verification report with confidence scores for every claim. // FAQ QUESTIONS ABOUT FACT CHECKING How does it verify claims? - The fact checker scans your blog post for statistical claims (numbers, percentages, dates, rankings). For each claim, it identifies the cited source URL, fetches the page content, and searches for the specific data point. It then compares the claim in your post against the source and produces a match confidence score. What is the confidence score? + The confidence score ranges from 0.0 to 1.0. A score of 1.0 means the exact number was found in the source. Scores between 0.7 and 0.99 indicate approximate matches where the source contains similar but not identical data. Scores below 0.5 suggest the claim could not be verified and needs manual review. Can it check any URL? + The fact checker can fetch and parse most public web pages. It handles HTML pages, PDFs, and plain text. Pages behind paywalls, login walls, or heavy JavaScript rendering may not be fully accessible. The checker reports when a source URL cannot be fetched so you can verify manually. Does it find replacement sources? + When a source URL returns a 404 or the claim cannot be verified, the fact checker suggests alternative source URLs where the same data point might be found. These suggestions are based on the claim context and common authoritative sources in the topic area. // RELATED SKILLS EXPLORE MORE BLOG ANALYZE Deep analysis of individual posts with readability context, evidence checks, and content-quality assessment. LEARN MORE → BLOG WRITE AI-powered blog post writing with SEO optimization, persona-aware tone, and automatic internal linking. LEARN MORE → BLOG REWRITE Rewrite and refresh existing blog posts with updated information, improved structure, and persona alignment. LEARN MORE → NOTEBOOKLM RESEARCH Source-grounded research from a curated corpus. Citations stay locked to your trusted documents, no model hallucination. NOTEBOOKLM RESEARCH → GOOGLE APIS Search Console, GA4, PageSpeed, CrUX, and Indexing API in one skill. Real ranking and traffic data, free quota. GOOGLE APIS → VIEW ALL SKILLS → VERIFY YOUR CLAIMS IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Keyword Cannibalization Detector | Claude Blog URL: https://claude-blog.md/skills/blog-cannibalization Keyword Cannibalization Detector | Claude Blog Skip to main content CLAUDE BLOG HOMESKILLSBLOG HOMESKILLSBLOGGITHUB Home/Skills/Keyword Overlap // SKILL KEYWORD CANNIBALIZATION Use this skill in the hosted vault Find blog posts that compete against each other. The detector extracts primary keywords from titles and headings, clusters semantically similar targets, and flags posts competing for the same search intent. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog cannibalization ./content/blog COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS FIND. CLUSTER. RESOLVE. Running /blog cannibalization scans all posts in your blog directory. It extracts the primary keyword target from each post's title, H1, and meta description, then uses semantic similarity to cluster posts targeting the same search intent. For each cluster, you get a recommendation: merge the content, redirect the weaker post, or differentiate the keyword targets. No more posts competing against themselves. 01 KEYWORD EXTRACTION Extracts primary and secondary keywords from each post's title, H1 heading, meta description, and H2 headings. Builds a keyword profile for every post in your blog. 02 SEMANTIC CLUSTERING Groups posts with semantically similar keyword targets. Uses text similarity scoring to identify posts that would compete for the same SERP position, even with slightly different wording. 03 INTENT MATCHING Classifies the search intent of each keyword target (informational, transactional, navigational). Posts with the same intent and similar keywords are the highest-risk cannibalization pairs. 04 MERGE RECOMMENDATIONS For each cannibalization cluster, recommends whether to merge content into a single comprehensive post, 301 redirect the weaker page, or differentiate by shifting keyword focus. // USAGE HOW TO DETECT OVERLAP Step 1: Install Claude Blog git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh Step 2: Open Claude Code claude Step 3: Run the Detector /blog cannibalization ./content/blog Point it at your blog directory. The detector scans all posts, clusters competing keyword targets, and outputs a report with merge/redirect/differentiate recommendations. // FAQ QUESTIONS ABOUT CANNIBALIZATION What is keyword cannibalization? - Keyword cannibalization occurs when multiple pages on your site target the same or very similar keywords. Search engines struggle to determine which page to rank, often resulting in both pages ranking lower than a single, comprehensive page would. It splits your authority and click-through rate across competing pages. How does it detect overlap? + The detector extracts primary keywords from each post's title, H1, H2 headings, and meta description. It then clusters posts with semantically similar keyword targets using text similarity analysis. Posts within the same cluster are flagged as potential cannibalization with a similarity score. Does it suggest which post to keep? + Yes. For each cannibalization cluster, the detector recommends a primary page based on content depth, word count, internal link count, and publication date. It then suggests actions for secondary pages: merge content into the primary, redirect with a 301, or differentiate by adjusting the target keyword. Can it use live SERP data? + The base detector works offline using your local blog files. If you have a DataForSEO API key configured, it can also check live SERP rankings to identify which of your pages currently rank for the same queries, providing real-world confirmation of cannibalization issues. // RELATED SKILLS EXPLORE MORE BLOG AUDIT Full-site blog health assessment with per-post quality scores, orphan detection, and stale content identification. LEARN MORE → BLOG STRATEGY Content strategy planning with topic clusters, competitive gaps, and publishing roadmaps. LEARN MORE → BLOG ANALYZE Deep analysis of individual posts with readability context, evidence checks, and content-quality assessment. LEARN MORE → TOPIC CLUSTERS SERP-overlap clustering, not text similarity. Hub-and-spoke architecture for topical authority. Community-built. TOPIC CLUSTERS → FLOW FRAMEWORK 30 evidence-led blogging prompts across Find, Optimize, Win stages. Structured prompts, not improvised queries. CC BY 4.0. FLOW FRAMEWORK → VIEW ALL SKILLS → DETECT OVERLAP IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## CMS Taxonomy Manager - Tags | Claude Blog URL: https://claude-blog.md/skills/blog-taxonomy CMS Taxonomy Manager - Tags | Claude Blog Skip to main content CLAUDE BLOG HOMESKILLSBLOG HOMESKILLSBLOGGITHUB Home/Skills/CMS Taxonomy // SKILL CMS TAXONOMY MANAGEMENT Use this skill in the hosted vault Manage tags and categories across your CMS. Supports WordPress REST API, Shopify GraphQL, Ghost Content API, Strapi REST/GraphQL, and Sanity GROQ. Suggest new tags from content analysis, sync taxonomy, and audit for orphan or duplicate tags. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog taxonomy audit --cms wordpress COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS TAGS. CATEGORIES. SYNCED. Running /blog taxonomy connects to your CMS API and pulls your full taxonomy (tags, categories, collections). It audits for orphan tags (no posts assigned), duplicate tags (similar names), and missing tags (topics in your content without corresponding tags). The manager also suggests new tags based on content analysis and can sync taxonomy changes back to your CMS with explicit confirmation before any write operation. 01 MULTI-CMS SUPPORT Works with WordPress (REST API v2), Shopify (Admin GraphQL), Ghost (Content/Admin API), Strapi (REST/GraphQL), and Sanity (GROQ). Each integration handles the platform's native taxonomy structure. 02 TAG SUGGESTIONS Analyzes your post content and suggests relevant existing tags or proposes new ones. Suggestions are ranked by relevance and matched against your current taxonomy to avoid duplicates. 03 TAXONOMY SYNC Push taxonomy changes to your CMS: create new tags, merge duplicates, update descriptions, and assign tags to posts. All write operations require explicit confirmation. 04 ORPHAN DETECTION Finds tags and categories with zero posts assigned. Orphan taxonomy items clutter your CMS and confuse navigation. The audit flags them for cleanup or reassignment. // USAGE HOW TO MANAGE TAXONOMY Step 1: Install Claude Blog git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh Step 2: Open Claude Code claude Step 3: Audit Your Taxonomy /blog taxonomy audit --cms wordpress Specify your CMS platform. The manager connects via API, pulls your taxonomy, and produces an audit report with orphan tags, duplicates, and missing tag suggestions. // FAQ QUESTIONS ABOUT TAXONOMY Which CMS platforms does it support? - The taxonomy manager supports five CMS platforms: WordPress (via REST API v2), Shopify (via Storefront and Admin GraphQL APIs), Ghost (via Content and Admin APIs), Strapi (via REST and GraphQL APIs), and Sanity (via GROQ queries and Mutations API). Each integration handles the platform's native taxonomy structure. Can it create tags automatically? + Yes. When configured with write access to your CMS API, the taxonomy manager can create new tags and categories directly. It can also assign tags to posts, update tag descriptions, and merge duplicate tags. All write operations require explicit confirmation before execution. How does tag suggestion work? + The tag suggester analyzes your post content, extracts key topics and entities, and compares them against your existing taxonomy. It suggests existing tags that match the content, and proposes new tags for topics not yet covered in your taxonomy. Suggestions are ranked by relevance. Does it handle categories too? + Yes. The taxonomy manager handles both tags and categories (or their equivalents on each platform). It manages hierarchical categories with parent-child relationships, flat tag lists, and platform-specific taxonomy types like Shopify collections or Ghost internal tags. // RELATED SKILLS EXPLORE MORE BLOG AUDIT Full-site blog health assessment with per-post quality scores, orphan detection, and stale content identification. LEARN MORE → SEO CHECKER Post-writing SEO validation with pass/fail checklist covering title tags, meta descriptions, and OG tags. LEARN MORE → BLOG STRATEGY Content strategy planning with topic clusters, competitive gaps, and publishing roadmaps. LEARN MORE → TOPIC CLUSTERS SERP-overlap clustering, not text similarity. Hub-and-spoke architecture for topical authority. Community-built. TOPIC CLUSTERS → MULTILINGUAL PUBLISHING One command, every language. Source post plus translated and localized variants plus hreflang in a single pipeline. MULTILINGUAL PUBLISHING → VIEW ALL SKILLS → MANAGE YOUR TAXONOMY IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## FLOW Framework: Evidence-Led Blogging Prompts | Claude Blog URL: https://claude-blog.md/skills/blog-flow FLOW Framework: Evidence-Led Blogging Prompts | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/FLOW Framework // SKILL FLOW FRAMEWORK EVIDENCE-LED BLOGGING Use this skill in the hosted vault Stop running improvised AI queries against your blog workflow. FLOW gives Claude Blog 30 evidence-led prompts organized across three stages: Find, Optimize, Win. Every analysis is anchored in a structured prompt with stated evidence, not vibes. Bundled under CC BY 4.0. Free with Claude Blog. FRAMEWORK © DANIEL AGRICI · CC BY 4.0 · github.com/AgriciDaniel/flow $ /blog flow [stage] COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS 3 STAGES. 30 PROMPTS. FLOW is an operating model for the AI-search era. Claude Blog exposes the three blog-relevant stages (Find, Optimize, Win) and keeps the single Leverage prompt available through the prompts index. Each stage ships with structured prompts that Claude Blog surfaces on demand, scoped to the situation you describe. Instead of guessing how to query an LLM about your content, you pick the stage you are in and FLOW selects the right prompts. The Optimize stage uses context-matching orchestration: with 21 prompts available, Claude Blog loads only the 2-3 most relevant based on your niche, URL signals, and prior skill output already in your conversation (like a recent /blog analyze score). F FIND Keyword discovery, intent mapping, gap analysis. 5 prompts. Use when starting a new topic and you do not yet know what to write. Hands off to /blog brief and /blog outline downstream. O OPTIMIZE 21 prompts covering on-page, technical, content quality, freshness, structured data. Claude Blog selects the 2-3 most relevant per analysis, never dumps all 21. Hands off to /blog rewrite, /blog seo-check, /blog geo, /blog schema, /blog factcheck. W WIN Bottom-of-funnel and conversion. 3 prompts. Use on money pages, product reviews, and pillar pages to convert organic traffic into pipeline. Hands off to /blog repurpose and /blog audio. Σ CONTEXT-MATCHED The Optimize stage never dumps all 21 prompts. It selects 2-3 based on niche, URL signals, and prior skill output (analyze score deltas, factcheck flags, geo readiness). Surface area without noise. // USAGE HOW TO RUN FLOW Stage Commands Command What it does /blog flow Shows the stage menu with one-line descriptions /blog flow find [topic|url] Runs all 5 Find prompts for keyword + intent discovery /blog flow optimize [url] Selects and runs 2-3 of 21 Optimize prompts by context /blog flow win [url] Runs all 3 Win prompts for BOFU and conversion /blog flow prompts Lists every prompt with stage, name, and trigger /blog flow sync Pulls the latest prompts from github.com/AgriciDaniel/flow Cross-Skill References FLOW does not duplicate the rest of Claude Blog. It points at the right specialist skill when one is needed: Find → /blog brief and /blog outline for downstream production Optimize → /blog rewrite, /blog seo-check, /blog geo, /blog schema, /blog factcheck Win → /blog repurpose and /blog audio for downstream distribution Why the Local Stage is Excluded The full FLOW framework includes 11 Local-stage prompts targeting Google Business Profile, NAP consistency, citations, and map-pack visibility. Those prompts target brick-and-mortar work, not blog content. For local SEO use the /seo local skill from the claude-seo plugin instead. Syncing the Prompt Library FLOW prompts live in the upstream repo at github.com/AgriciDaniel/flow. Run /blog flow sync to pull the latest version. The sync script is anonymous-first against the GitHub API (no token required for under 60 requests per hour on public repos). A SHA-256 lockfile (flow-prompts.lock) tracks baseline integrity and reports upstream drift before writing. Attribution Every /blog flow activation outputs this header before analysis: Framework and prompts © Daniel Agrici, CC BY 4.0, github.com/AgriciDaniel/flow The attribution is required by the CC BY 4.0 license and is automated by the sync script. Do not remove it. // FAQ QUESTIONS ABOUT FLOW What is the FLOW framework for bloggers? - FLOW is an evidence-led operating model built for the AI-search era. For blogs it organizes work into three stages: Find (keyword discovery, intent mapping, gap analysis), Optimize (on-page, technical, content quality, freshness, structured data), and Win (BOFU, conversion, dual-surface scorecard). Each stage ships with structured AI prompts that drive analysis based on evidence rather than improvised queries. How many prompts does the FLOW library include? + 30 blog-applicable prompts total: 5 Find prompts, 21 Optimize prompts, and 3 Win prompts, plus 1 Leverage prompt accessible via the full prompts index. The Optimize stage uses context-matching orchestration: instead of dumping all 21 prompts, Claude Blog selects 2-3 most relevant based on niche, URL signals, and any prior skill output (like a recent /blog analyze score) in the conversation. Is the FLOW prompt library open-source? + Yes. The FLOW prompts are released under CC BY 4.0 by Daniel Agrici and bundled inside Claude Blog with attribution. Claude Blog's skill code remains MIT-licensed. You can use, modify, and redistribute the prompts as long as attribution to the FLOW repo (github.com/AgriciDaniel/flow) is preserved. How does FLOW integrate with the rest of Claude Blog? + FLOW points to the right specialist skill at every stage rather than duplicating their logic. Find stage hands off to /blog brief and /blog outline. Optimize stage hands off to /blog rewrite, /blog seo-check, /blog geo, /blog schema, and /blog factcheck. Win stage hands off to /blog repurpose and /blog audio. The framework sits on top of the existing 29 skills rather than replacing any of them. Why is the Local stage excluded for blogs? + The full FLOW framework includes 11 Local-stage prompts targeting Google Business Profile, NAP consistency, citations, and map-pack visibility. Those prompts are intentionally excluded from blog-flow because they target brick-and-mortar local SEO, not blog content workflows. For local SEO use /seo local from the claude-seo plugin instead. // RELATED SKILLS EXPLORE MORE BLOG STRATEGY Niche positioning, topic ideation, and content pillar architecture. FLOW Find stage hands off to strategy for execution planning. BLOG STRATEGY PLANNER → TOPIC CLUSTERS SERP-overlap-based semantic clustering. FLOW Find stage hands off to cluster for hub-and-spoke architecture turning discovery into structure. SEMANTIC CLUSTERING → CONTENT REWRITER FLOW Optimize stage selects 2-3 relevant prompts; /blog rewrite applies them to your existing posts with freshness updates. CONTENT OPTIMIZER → VIEW ALL 30 SKILLS → RUN YOUR FIRST FLOW LOOP IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Google APIs for Blogs: GSC, PageSpeed, GA4 | Claude Blog URL: https://claude-blog.md/skills/blog-google Google APIs for Blogs: GSC, PageSpeed, GA4 | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Google APIs // SKILL GOOGLE APIS FOR BLOGGERS Use this skill in the hosted vault Direct access to Google's own data for blog content workflows. Real Chrome user metrics from CrUX, real search performance from Search Console, real GA4 organic traffic, plus PageSpeed Insights for Core Web Vitals. All APIs are free. Setup walks you through Google Cloud project + OAuth in 5 minutes. 13 commands across 4 credential tiers. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog google pagespeed COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // WHAT YOU GET REAL GOOGLE DATA. NO SCRAPING. Claude Blog connects to Google's APIs for first-party and official data. PageSpeed Insights can return CrUX field data from real Chrome users. Search Console returns clicks, impressions, CTR, and position. GA4 returns your organic traffic. The Indexing API is restricted to eligible job-posting and livestream pages, so ordinary blog pages use URL Inspection, sitemaps, internal links, and quota-limited Request Indexing for important URLs. The skill detects configured credentials and gates commands by tier. Start with an API key for PageSpeed, CrUX, YouTube, and NLP. Add OAuth or a service account for Search Console, URL Inspection, and sitemaps. Indexing API commands remain eligibility-gated. GA4 and Ads credentials unlock their respective reporting tools. 01 PAGESPEED + CRUX Combined Lighthouse lab data and real Chrome user metrics. p75 LCP, INP, CLS, FCP, TTFB. 25-week history for trend analysis. 02 SEARCH CONSOLE Clicks, impressions, CTR, position, top queries, top pages. URL Inspection for crawl status. Sitemap submission monitoring. 03 GA4 ORGANIC Organic traffic, top landing pages, engagement, conversions. Pulls last 28 days by default. Requires GA4 property ID configured. 04 INDEXING + NLP Eligibility-gated Indexing API support for permitted page types. Ordinary posts use inspection and sitemap workflows. NLP adds entity and sentiment context. // USAGE 13 COMMANDS. 4 TIERS. Command Reference Command Description Tier /blog google setup Configure API credentials -- /blog google pagespeed PSI + CrUX field data 0 /blog google crux CrUX field data only 0 /blog google crux-history 25-week CWV trends 0 /blog google gsc Search Console metrics 1 /blog google inspect URL Inspection 1 /blog google index Eligibility-gated Indexing API submit 1 /blog google ga4 [property-id] GA4 organic traffic 2 /blog google ga4-pages [property-id] Top organic landing 2 /blog google youtube YouTube video search 0 /blog google nlp Entity extraction + sentiment 0 /blog google keywords Google Ads Keyword Planner 3 Credential Tiers Tier 0 (API key): PageSpeed, CrUX, CrUX history, YouTube, NLP. Free API key from Google Cloud Console, no OAuth flow. Tier 1 (OAuth or service account): Adds Search Console, URL Inspection, sitemaps, and eligibility-gated Indexing API support. Tier 2 (GA4 property ID): Adds GA4 organic traffic and top landing pages. Requires a GA4 property already collecting data. Tier 3 (Ads developer token): Adds Google Ads Keyword Planner. Requires an approved Google Ads developer token and a customer ID. Claude Blog auto-detects your tier from the config file and only surfaces commands you can actually run. Run /blog google setup first to walk through Tier 0 and 1 in five minutes. Sharing config with claude-seo blog-google and seo-google read the same config file at ~/.config/claude-seo/google-api.json. If you have both plugins installed, you only set up credentials once. Run /seo google setup or /blog google setup and the OAuth tokens, GA4 property ID, and Ads tokens are shared. The two skills do not duplicate functionality: claude-seo targets site-wide audits while claude-blog targets blog content workflows. Config File Schema { "api_key": "AIzaSy...", "oauth_client_path": "/path/to/client_secret.json", "default_property": "sc-domain:example.com", "ga4_property_id": "properties/123456789", "ads_developer_token": "...", "ads_customer_id": "123-456-7890" } All fields are optional. The presence of each field promotes you to the next tier. // FAQ QUESTIONS ABOUT GOOGLE APIS Do I need to pay for Google APIs? - Many supported APIs have free access or quotas, but account, billing, eligibility, and quota requirements vary. Ads access and some Google Cloud services can have additional requirements. Check the current official documentation before enabling a service. What is the 4-tier credential system? + Tier 0 covers API-key tools such as PageSpeed, CrUX, YouTube, and NLP. Tier 1 adds Search Console, URL Inspection, and sitemaps. Indexing API support is restricted to eligible page types. Tier 2 adds GA4 and Tier 3 adds Ads Keyword Planner. How does blog-google integrate with claude-seo? + Both skills can share the same local credentials configuration. Claude SEO focuses on site-wide audits, while Claude Blog applies the data to blog workflows. What can I do with NLP entity extraction? + Run /blog google nlp on a URL or text to inspect entities, sentiment, and classification. Treat these outputs as analytical context, not as direct E-E-A-T or ranking-factor scores. How does the 25-week CrUX history help? + CrUX history shows how field performance metrics changed over time when sufficient data is available. It can help identify regressions and evaluate performance work alongside deployment history. // RELATED SKILLS EXPLORE MORE AI CITATION OPTIMIZER Pair Google API CrUX data with GEO scoring to validate that fast Core Web Vitals correlate with AI search citations. AI CITATION OPTIMIZER → BLOG ANALYZE Blog analyze scoring feeds CrUX + GSC data for objective ranking metrics. Quality score plus real Google data plus URL inspection. BLOG ANALYZE → FLOW FRAMEWORK FLOW's Optimize stage uses CrUX field data and GSC performance to select 2-3 relevant optimize prompts. Evidence-led. FLOW FRAMEWORK → VIEW ALL 30 SKILLS → PULL REAL GOOGLE DATA IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## NotebookLM Research for Bloggers | Claude Blog URL: https://claude-blog.md/skills/blog-notebooklm NotebookLM Research for Bloggers | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/NotebookLM Research // SKILL NOTEBOOKLM RESEARCH SOURCE-GROUNDED CITATIONS Use this skill in the hosted vault Stop asking general LLMs questions and getting plausible-sounding answers without sources. NotebookLM lets Claude Blog query your own curated knowledge base, returning citation-ready answers with exact source attribution. Drop research PDFs, web articles, and meeting notes into a NotebookLM notebook; ask questions in Claude Blog; get back claims with file-and-page citations you can fact-check. Browser OAuth flow, free to use. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog notebooklm COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS YOUR DOCS. CITED ANSWERS. NotebookLM is Google's source-grounded research tool. You upload a corpus of documents (PDFs, Drive files, web URLs, YouTube transcripts) into a notebook, then ask questions answered exclusively from those sources. Every answer ships with file and page citations you can verify before quoting in your blog. Claude Blog wraps this workflow so you never leave the terminal. Run /blog notebooklm setup once to authenticate via Google OAuth, then query any notebook by name or URL during your writing flow. Pair it with /blog write to seed drafts with research that survives fact-checking. 01 SOURCE-GROUNDED Every answer includes file-and-page citations from your own notebooks. No more 'hallucinated' research; every claim traces back to a real document. 02 OWN YOUR KNOWLEDGE BASE Build a notebook of trusted sources: research papers, industry reports, internal docs, transcripts. NotebookLM only answers from your curated corpus. 03 BROWSER OAUTH FLOW Authenticate once via your Google account; Claude Blog handles session management. No API key required. Free to use. 04 BLOG RESEARCH WORKFLOW Drop a /blog write topic; query /blog notebooklm for source-grounded background; cite findings inline in your draft with full attribution preserved. // USAGE HOW TO QUERY NOTEBOOKLM Commands Command What it does /blog notebooklm Queries the active notebook and returns a source-grounded answer with file and page citations /blog notebooklm setup Runs the one-time Google OAuth flow in a browser tab and caches the session locally Building a research notebook NotebookLM accepts PDFs, Google Drive docs, web URLs, and YouTube transcripts as source inputs. The skill quality is bounded by the corpus quality, so curate deliberately. A typical topic notebook benefits from 10 to 30 trusted sources: peer-reviewed papers for the empirical claims, industry reports for benchmarks, primary documentation for technical specs, and a handful of expert blog posts for framing. Once your notebook is populated, you can query it from Claude Blog with natural language. NotebookLM only answers from the uploaded corpus, so a sparse or off-topic notebook will produce sparse or off-topic answers. Treat the notebook as the research substrate, not the question scope. Inside the writing loop The typical flow: kick off /blog write with a topic; pause when you need a citable statistic; run /blog notebooklm against your topic notebook; paste the returned passage with its citation back into your draft. /blog factcheck later verifies the numbers against the same sources, so the trail stays clean from research to publish. // FAQ QUESTIONS ABOUT NOTEBOOKLM What is NotebookLM? - Google NotebookLM is a free research tool that lets you upload PDFs, Drive docs, web URLs, and YouTube transcripts into a notebook, then ask questions answered only from those sources. Unlike general LLMs, every answer is source-grounded with file and page citations. Claude Blog wraps NotebookLM so you can query notebooks from your terminal during blog writing workflows. Is NotebookLM free? + Yes. NotebookLM has a free tier sufficient for most blogging workflows. You can create unlimited notebooks with up to 50 sources each. Premium tiers exist for higher source limits and team features but are not required to use this skill. How is this different from general LLM research? + General LLMs answer from their training data, which is a moving target with no citation. NotebookLM answers only from sources you uploaded, with file and page citations. For evidence-based blogging this is the difference between citable research and 'I think I read somewhere that...'. Can I cite NotebookLM in my blog post? + Indirectly. NotebookLM gives you citations to the underlying sources; cite those sources directly in your blog post, not NotebookLM itself. Use NotebookLM as a research efficiency tool to surface relevant passages from your curated corpus quickly. How does setup work? + Run /blog notebooklm setup. Claude Blog opens a browser tab for Google OAuth, you grant access to your NotebookLM, and the session is cached locally. Re-auth happens automatically on token refresh. No API key, no service account configuration needed. // RELATED SKILLS EXPLORE MORE FACT CHECKER NotebookLM finds source-grounded answers; /blog factcheck verifies the statistics in your draft against those same sources. FACT CHECKER → CONTENT BRIEFS Run NotebookLM during brief generation to ground your topic in trusted research before you start writing. CONTENT BRIEFS → FLOW FIND STAGE FLOW's Find stage hands off to NotebookLM for source-grounded research after keyword discovery and intent mapping. FLOW FIND STAGE → VIEW ALL 30 SKILLS → START SOURCE-GROUNDED RESEARCH IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## AI Audio Narration: 30 Voices, 80+ Languages | Claude Blog URL: https://claude-blog.md/skills/blog-audio AI Audio Narration: 30 Voices, 80+ Languages | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Audio Narration // SKILL AUDIO NARRATION 30 VOICES, 80+ LANGUAGES Use this skill in the hosted vault Every blog post has a podcast audience that will never read the words. Audio narration makes your content accessible to commuters, screen-readers, and accessibility tools without re-recording. Claude Blog wraps Google Gemini TTS to generate broadcast-quality narration directly from your published markdown. 30 voices across 80+ languages, multi-speaker dialogue support, 24kHz mono WAV output. Free Google AI Studio API key, free output. BY DANIEL AGRICI · UPDATED MAY 2026 $ /blog audio generate COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS PODCAST-QUALITY AUDIO FROM MARKDOWN. Audio narration is the simplest accessibility upgrade your blog can ship. Screen readers handle the basic case, but a real human-sounding narrator opens your content to a completely different audience: people who would rather listen during a commute, a workout, or while cooking dinner. Claude Blog wraps Google Gemini TTS so you can generate that narration directly from the same markdown that powers your post. Pick a voice, run /blog audio generate, get a 24kHz mono WAV back. No re-recording, no DAW, no studio booking. The generator handles multi-speaker dialogue automatically when your post contains speaker tags, which makes interview-format and debate-format posts work without extra effort. 01 30 GEMINI VOICES Charon, Kore, Puck, Aoede, Fenrir, and 25 more. Each voice has a distinct timbre. Preview voices with /blog audio voices before committing. 02 80+ LANGUAGES Generate narration in any major language. Multilingual posts can output one audio file per language with native pronunciation. 03 MULTI-SPEAKER DIALOGUE Interview-style content supports multiple voices per file. Annotate dialogue with speaker tags; each speaker gets a distinct voice automatically. 04 24KHZ MONO WAV Broadcast-quality output. 24kHz mono WAV format, suitable for podcast distribution, in-page audio players, or accessibility playback. // USAGE HOW TO RUN AUDIO Commands Command What it does /blog audio generate Generates a narration audio file from a blog markdown post /blog audio voices Lists all 30 available voices with sample audio previews /blog audio setup Configures the Google AI Studio API key for Gemini TTS Voice selection Voices map to tones. Charon is authoritative and works well for analytical or technical posts. Kore is warm and suits storytelling, lifestyle, or interview content. Puck is energetic and works for upbeat marketing pieces, product launches, and high-energy explainers. Aoede is breezy and reads well for casual narratives. Fenrir is deeper and grounded, useful for serious analysis. Each voice carries the same multilingual range, so pick the timbre that matches your post, not the language. Costs and limits Gemini TTS is free at the Google AI Studio level for normal blog cadence. Rate limits apply (requests per minute, characters per day), which is fine for a typical publishing rhythm of 1 to 3 posts per week. For podcast-scale workflows or full-archive backfills, the paid Gemini API tier is billed per million characters of input text at premium pricing. Output audio itself is not metered, only the input text is counted. // FAQ QUESTIONS ABOUT AUDIO Is Gemini TTS free? - Yes for normal blog usage. The Google AI Studio free tier covers tens of thousands of characters per day, sufficient for a blog cadence of 1-3 posts per week. Premium tiers (Gemini API paid) lift the limit for podcast-scale workflows. Pricing currently around $10 per million characters at the paid tier. Which voices are available? + 30 voices including Charon, Kore, Puck, Aoede, Fenrir, Sulafat, Algieba, Despina, Erinome, Schedar, Umbriel, Algenib, Rasalgethi, Laomedeia, Achernar, Alnilam, Achird, Zubenelgenubi, Vindemiatrix, Sadachbia, Sadaltager, Iapetus, Gacrux, Pulcherrima, Enceladus, and more. Run /blog audio voices to preview each one. Can I do multi-speaker dialogue? + Yes. Annotate dialogue in your blog markdown with speaker tags like 'Alice:' and 'Bob:'. The audio generator detects speaker shifts and assigns distinct voices automatically. Useful for interview content, debate-format posts, and dialogue-driven explainers. What languages are supported? + 80+ languages including English, Spanish, French, German, Japanese, Korean, Mandarin, Portuguese, Italian, Dutch, Russian, Arabic, Hindi, Indonesian, Vietnamese, Thai, Turkish, Polish, Swedish, Danish, and many more. Native pronunciation is automatic from the source text language detection. What format is the output? + 24kHz mono WAV by default. The output is broadcast-quality, suitable for podcast distribution, in-page HTML audio players, or accessibility playback tools. Convert to MP3 or AAC for distribution if needed. // RELATED SKILLS EXPLORE MORE BLOG REPURPOSE Audio is one of the repurposing surfaces. Pair audio narration with /blog repurpose to ship the same post as text, audio, social cards, and newsletter. CONTENT REPURPOSING → MULTILINGUAL PUBLISHING Generate narration in every language your post was translated to. One source markdown, multiple audio outputs. MULTILINGUAL BLOG → FLOW WIN STAGE FLOW's Win stage hands off to /blog audio for distribution: turn your money pages into audio to capture commute listening. FLOW FRAMEWORK → VIEW ALL 30 SKILLS → ADD AUDIO TO YOUR BLOG IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Topic Cluster Engine: SERP-Based Content Hub | Claude Blog URL: https://claude-blog.md/skills/blog-cluster Topic Cluster Engine: SERP-Based Content Hub | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Topic Clusters // SKILL TOPIC CLUSTER ENGINE SERP-BASED HUB AND SPOKE Use this skill in the hosted vault Most topic clustering tools group keywords that sound alike. That misses what actually matters: which keywords Google ranks for the same pages. blog-cluster groups by actual SERP overlap, designs a hub-and-spoke content architecture, and outputs an interactive cluster map plus automatic internal-link injection across the cluster's posts. If you have /blog write installed, the cluster can also be executed end-to-end. CONTRIBUTED BY LUTFIYA MILLER · PRO HUB CHALLENGE WINNER · contributor attribution $ /blog cluster plan COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS FOUR LAYERS. ONE CLUSTER. blog-cluster fills the gap between strategy and execution. blog-strategy plans the blueprint, blog-write produces one post, blog-cluster builds the entire interlinked house. It starts from a single seed keyword, expands the keyword universe via search, groups by SERP overlap, designs the hub-and-spoke architecture, and outputs the cluster map plus the recommended internal linking matrix. The clustering signal is the key. Two keywords belong in the same cluster only if Google ranks substantially the same URLs for both. Text similarity is not enough. The map renders as a static HTML file with inline SVG, sized by total search volume per cluster, and hardened against XSS during integration into Claude Blog. 01 SERP-OVERLAP CLUSTERING Measures actual Google SERP overlap between keywords. Two keywords are in the same cluster only if Google ranks substantially the same URLs for both. 02 HUB AND SPOKE ARCHITECTURE Designs each cluster as a pillar page (hub) plus 3-12 supporting articles (spokes). Suggests URLs, slugs, and content templates per intent. 03 AUTOMATIC INTERNAL LINKING Generates the recommended internal linking matrix: hub-to-spoke, spoke-to-spoke, sibling clusters. Anchor text suggestions per link. Auto-injected by /blog cluster execute. 04 INTERACTIVE CLUSTER MAP Outputs cluster-map.html sized by total search volume. Drop in browser, no build step. XSS-hardened during integration (no user-controlled HTML injection). // USAGE HOW TO RUN BLOG CLUSTER Commands Command What it does /blog cluster plan SERP-based semantic analysis from a seed keyword. Outputs cluster-plan.json and cluster-map.html with hub, spokes, intent classification, and the internal linking matrix. /blog cluster execute Sequential per-post writes using the plan. Hands off to /blog write for the pillar and every spoke, with cluster context and auto-injected interlinks. /blog cluster map Regenerates the interactive cluster-map.html from an existing plan file. Useful after manual edits to the plan. How SERP-overlap scoring works For each pair of candidate keywords, blog-cluster pulls the top 10 ranking URLs from Google for each keyword and computes a Jaccard-style similarity over the two URL sets. The default similarity threshold for cluster inclusion is 30%, which corresponds to at least 3 of 10 shared results between any two keywords in a cluster. Below the threshold, keywords split into separate clusters. Above it, they collapse into one. The threshold is tunable; a higher value produces tighter clusters and more pages, a lower value produces broader clusters and fewer pages. Execution path If /blog write is installed, /blog cluster execute writes the hub and every spoke with the agreed internal linking matrix injected during draft generation. Each spoke knows about the hub and its siblings before it is written, so anchor text and internal links are placed naturally inside the prose rather than appended as a related-posts list. If /blog write is not installed, the skill outputs an editorial brief per post with a title, primary topic, intent, template, intent-dependent coverage, and contextual internal-link suggestions. The brief can be handed to a writer or fed back into any LLM with no further dependencies. Output artifacts cluster-plan.json: full machine-readable plan with pillar, clusters, posts, link matrix, and volume estimates. cluster-map.html: static HTML with inline SVG. No JavaScript. Hover and tooltips use CSS and native SVG only. pillar-<slug>.md and <spoke-slug>.md: per-post markdown drafts (Execute Phase, requires /blog write). cluster-scorecard.md: per-post status, cluster cohesion score, link audit, cannibalization check. Attribution The original semantic-cluster-engine submission by Lutfiya Miller scored 95/100 Exemplary in the Pro Hub Challenge audit (security, functionality, code quality, documentation, innovation). Lutfiya's Plan + Execute architecture and the cluster-context injection pattern are preserved verbatim. The Claude Blog port removes brand-specific styling and image prompts from the original, hardens the cluster-map HTML output against XSS, and routes through existing claude-blog sub-skills (/blog write, /blog strategy, /blog schema). The contribution is documented in Claude Blog contributor credits. // FAQ QUESTIONS ABOUT BLOG CLUSTER Why SERP overlap instead of text similarity? - Text similarity groups keywords that sound alike. SERP overlap groups keywords Google treats as the same intent. They are not the same. 'best running shoes' and 'top running sneakers' have low text similarity but high SERP overlap (Google ranks the same pages for both). Building a cluster on SERP overlap means one page can win the whole cluster. What does the cluster map show? + An interactive HTML file (cluster-map.html) with each cluster as a node, sized by total search volume. Edges connect related clusters. Each cluster expands to show its keywords, suggested URL, intent classification, and recommended internal links. Drop the file in any browser; no build step required. XSS-hardened during integration. Can Claude Blog actually write the cluster content? + Yes. With /blog write installed, /blog cluster execute hands off to write the hub and spokes with the agreed internal linking matrix. Without /blog write installed, the skill outputs an editorial brief you can hand to a writer or run through any LLM. How is intent classified? + By analyzing the SERP composition. Informational SERPs feature articles and guides. Commercial SERPs feature comparisons and listicles. Transactional SERPs feature product pages. Navigational SERPs feature a single brand. Each cluster gets one primary intent that drives URL pattern and template selection. Who contributed this skill? + Lutfiya Miller, winner of the AI Marketing Hub Pro Hub Challenge. The original contribution is documented in Claude Blog contributor credits. Lutfiya scored 95/100 Exemplary in the security and quality review. The submission was clean-room re-implemented for Claude Blog v1.7.0 with XSS hardening on the cluster-map output. Lutfiya is the same contributor who landed seo-cluster in claude-seo v1.9.0. // RELATED SKILLS EXPLORE MORE FLOW FIND STAGE FLOW's Find stage hands off to /blog cluster to turn discovered keywords into hub-and-spoke architecture. FLOW FRAMEWORK → BLOG STRATEGY Cluster output feeds strategy: niche positioning + topic pillars + cluster architecture is the full strategic stack. BLOG STRATEGY PLANNER → AI BLOG WRITER Cluster execute hands off to /blog write to produce the hub and spokes with internal links auto-injected. AI BLOG WRITER → VIEW ALL 30 SKILLS → BUILD YOUR FIRST CONTENT CLUSTER IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Multilingual Blog Publishing: One Command | Claude Blog URL: https://claude-blog.md/skills/blog-multilingual Multilingual Blog Publishing: One Command | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Multilingual Publishing // SKILL MULTILINGUAL PUBLISHING ONE COMMAND, EVERY LANGUAGE Use this skill in the hosted vault Writing in English and translating later is a friction-stack: you maintain four separate workflows, hreflang gets forgotten, cultural adaptation happens never. blog-multilingual collapses the stack into one command. Specify your topic plus a list of target languages; Claude Blog writes the source post, produces SEO-optimized translations with format preservation, runs cultural localization per locale, and emits hreflang tags ready for deploy. Four skills cooperating: write, translate, localize, locale-audit. CONTRIBUTED BY CHRIS MUELLER · PRO HUB CHALLENGE · original repo $ /blog multilingual <topic> --languages es,fr,de,ja COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS FOUR SKILLS. ONE PIPELINE. blog-multilingual is not a single skill but an orchestrator. It calls four cooperating skills in sequence: write (English source), translate (SEO-optimized format-preserving translation), localize (cultural adaptation per profile), and locale-audit (multilingual QA). At the end it emits hreflang tag stubs for every language pair plus x-default. The pipeline runs unattended once you specify languages. You get back a complete international content suite ready to deploy: source post plus N translated and localized variants, plus hreflang tags ready to paste into your CMS or static site generator. 01 ONE COMMAND /blog multilingual --languages es,fr,de,ja produces a complete international content suite. Source post plus 4 translated and localized variants, plus hreflang. 02 SEO-PRESERVED TRANSLATION Translations preserve markdown structure, frontmatter, schema blocks, inline SVG, and locale-correct number, date, currency, quote formatting. 03 CULTURAL ADAPTATION Beyond translation: DACH, Francophone, Hispanic, Japanese profiles adapt examples, idioms, and references to the target culture. Custom profiles supported. 04 HREFLANG AUTOGENERATED Outputs ready-to-deploy hreflang tags linking every language variant. No manual hreflang maintenance, no broken international targeting. // USAGE HOW TO RUN MULTILINGUAL Commands Command What it does /blog multilingual <topic> --languages <codes> Writes the source, translates, localizes, and emits hreflang for every target language /blog multilingual <topic> --languages es,fr,de,ja Concrete four-locale run. Produces source plus 4 translated and localized variants plus hreflang /blog multilingual <topic> --languages de,fr --source en --no-localize Translation only, cultural adaptation skipped. Useful when you have an existing brand voice you do not want re-styled /blog multilingual <topic> --languages es,ja --format mdx Force MDX output for Next.js, Astro, or Docusaurus targets. Defaults to auto-detect from your project Locale codes follow BCP 47. Plain ISO 639-1 codes (es, fr, de, ja) default to the most widely used regional variant; specify the full tag (es-MX, pt-BR, zh-TW) to pin a specific region. Cross-Skill References blog-multilingual delegates the heavy lifting to specialist skills. You can also invoke each one standalone: /blog translate for SEO-optimized translation of an existing file /blog localize for cultural adaptation of an already-translated file /blog locale-audit for post-publish multilingual QA across the whole suite Cultural Profiles Four built-in cultural profiles ship with the skill: DACH for German-Austrian-Swiss with formal/informal register awareness Francophone for French, Quebec French, and African French variants Hispanic for Spain, Mexico, and Argentina variants Japanese with formality registers (casual, polite, honorific) Custom profiles can be added via YAML configs in the skill's references directory. Each profile defines example substitutions, idiom mappings, legal-reference swaps, and CTA tone adjustments applied during the localize phase. // FAQ QUESTIONS ABOUT MULTILINGUAL How is this different from running translate four times? - blog-multilingual orchestrates four skills in sequence: write (English source), translate (SEO-optimized translation), localize (cultural adaptation), locale-audit (multilingual QA). Each step builds on the previous. Running translate alone gives you mechanical translations; running localize alone has nothing to localize. The orchestrator handles the full pipeline plus hreflang tag emission. What formats are preserved in translation? + Markdown, MDX, HTML, frontmatter YAML, JSON-LD schema blocks, inline SVG, code blocks, and locale-correct number, date, currency, quote formatting. The blog-translator agent (no Bash, per security best practice) handles format preservation while the translation happens. Which cultural profiles ship? + DACH (German-Austrian-Swiss with formal/informal register awareness), Francophone (French, Quebec French, African French variants), Hispanic (Spain, Mexico, Argentina variants), and Japanese (with formality registers: casual, polite, honorific). Custom profiles can be added via YAML configs in the skill's references directory. How are hreflang tags generated? + After write + translate + localize complete, the orchestrator emits hreflang tag stubs for every language pair, plus x-default. The output is ready-to-paste into your CMS or static site generator. /blog locale-audit verifies hreflang correctness across the full multilingual suite post-publish. Who contributed this skill? + Chris Mueller via the AI Marketing Hub Pro Hub Challenge. Chris also contributed seo-hreflang cultural profile enhancements to claude-seo v1.9.0. The multilingual blog suite scored 85/100 Proficient in the security and quality review. Original at github.com/Chriss54/claude-blog-multilingual. Integrated into Claude Blog v1.7.0 with Chris's permission. // RELATED SKILLS EXPLORE MORE SEO TRANSLATION Pure translation skill, SEO-optimized format preservation. Use standalone or via /blog multilingual. SEO TRANSLATION → CULTURAL LOCALIZATION Deep adaptation beyond translation: idioms, examples, references per cultural profile. CULTURAL LOCALIZATION → MULTILINGUAL AUDIT Post-publish QA across the full multilingual suite: completeness, hreflang correctness, meta parity, freshness. MULTILINGUAL AUDIT → VIEW ALL 30 SKILLS → PUBLISH IN 4 LANGUAGES IN ONE COMMAND. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## SEO Blog Translation: Format-Preserving | Claude Blog URL: https://claude-blog.md/skills/blog-translate SEO Blog Translation: Format-Preserving | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/SEO Translation // SKILL SEO BLOG TRANSLATION FORMAT-PRESERVING Use this skill in the hosted vault Generic LLM translation breaks SEO. Frontmatter gets reformatted, markdown structure drifts, schema blocks lose syntax validity, numbers and dates render in the source locale, quote marks come out as the wrong character. blog-translate solves all of that. It produces native-quality SEO translations while preserving every structural and locale-sensitive element of your post: markdown, MDX, HTML, frontmatter YAML, JSON-LD schema, inline SVG, code blocks, plus locale-correct number, date, currency, and quote formatting. CONTRIBUTED BY CHRIS MUELLER · PRO HUB CHALLENGE · original repo $ /blog translate <file> --to es,fr,de COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // WHY IT MATTERS FOUR REASONS TO USE IT Generic translation tools optimize for raw text fidelity. Blog publishing needs more than that. A blog post in production is a structured artifact: frontmatter that drives the build, JSON-LD schema that drives rich results, inline SVG that renders data visualizations, code blocks that must stay syntactically valid, and locale-sensitive characters that look right or wrong instantly to a native reader. blog-translate is built for that artifact, not just the prose inside it. The blog-translator agent runs without Bash and without untrusted-fetch execution, hardened against prompt injection per the v1.7.1 security baseline. Every translation is structurally identical to the source and locale-correct in every detail a search engine and a human reader will notice. 01 FORMAT PRESERVED Markdown structure, frontmatter YAML, JSON-LD schema, inline SVG, code blocks all preserved. The translation file is a structurally identical replacement, not a free-form output. 02 LOCALE-CORRECT FORMATTING Numbers (1,234.56 vs 1.234,56), dates (MDY vs DMY vs YMD), currency symbols, quote marks ("" vs „“ vs « »). Per-locale rules applied automatically. 03 SEO SIGNALS PRESERVED Title length stays within range. Meta description stays within range. Heading hierarchy preserved. Internal anchor structure mapped to the target language. 04 HARDENED AGENT Powered by the blog-translator agent. No Bash tool grant, no untrusted-fetch execution, prompt-injection hardened per the v1.7.1 security audit baseline. // USAGE HOW TO RUN IT Command What it does /blog translate <file> --to <codes> Translates the file into every comma-separated target locale, preserving format and applying locale rules Supported locales BCP 47 locale codes with regional variants are supported. A representative subset: Spanish: es-ES, es-MX, es-AR French: fr-FR, fr-CA German: de-DE, de-AT, de-CH East Asian: ja-JP, ko-KR, zh-CN, zh-TW Portuguese: pt-BR, pt-PT European: it-IT, nl-NL, ru-RU, tr-TR Other: ar-SA, hi-IN Plain ISO 639-1 codes (es, fr, de, ja) default to the most widely used regional variant. Quality bar Output is held to the same standard as source content. Every translated post passes the same /blog analyze 5-category scoring (Content, SEO, E-E-A-T, Technical, AI Citation). A translated post that drops a category score below the source baseline is flagged for re-pass before delivery. // FAQ COMMON QUESTIONS How is this different from Google Translate or DeepL? - Google Translate and DeepL produce raw text translations. They break markdown, drop frontmatter, mangle schema blocks, and use source-locale formatting for numbers and dates. blog-translate is built for blog publishing: it preserves every structural element, applies target-locale formatting rules, and maintains SEO signal integrity (title and meta length, heading hierarchy, internal anchor structure). What file formats are supported? + Markdown (.md), MDX (.mdx), HTML (.html), plus mixed-format files. The skill detects format from the file extension and content. Frontmatter (YAML or TOML) is parsed and translated field-by-field; structural keys (slug, date, image) are preserved as-is. Does it translate schema markup? + Yes, but selectively. Translatable schema fields (headline, description, articleSection) get translated. Structural fields (datePublished, url, @type) are preserved. Image and video URLs are not modified. The output schema is valid JSON-LD that passes Google Rich Results testing. Can I use custom locale profiles? + Yes. The 4 default cultural profiles (DACH, Francophone, Hispanic, Japanese) can be extended with YAML configs in the skill's references directory. Custom profiles can specify glossaries (preferred translations for technical terms), formality registers, idiom adaptations, and locale-specific examples to substitute. Who contributed this skill? + Chris Mueller via the AI Marketing Hub Pro Hub Challenge. Chris's multilingual contribution to claude-blog (blog-translate plus blog-multilingual, blog-localize, blog-locale-audit, blog-translator agent) scored 85/100 Proficient. Original at github.com/Chriss54/claude-blog-multilingual. Chris also contributed seo-hreflang cultural profile enhancements to claude-seo v1.9.0. // RELATED SKILLS EXPLORE MORE MULTILINGUAL ORCHESTRATOR One-command multilingual publishing. Use /blog multilingual for end-to-end international workflows; /blog translate as the translation step alone. MULTILINGUAL PIPELINE → CULTURAL LOCALIZATION Translate handles formats and locale formatting; localize handles cultural adaptation (idioms, examples, references). Pair for native-feeling content. CULTURAL LOCALIZE → MULTILINGUAL QA After translating, run locale-audit to verify hreflang correctness, meta parity, and freshness across all language variants. LOCALE AUDIT → VIEW ALL 30 SKILLS → TRANSLATE WITHOUT BREAKING SEO IN ONE COMMAND. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Cultural Localization: DACH, FR, ES, JA | Claude Blog URL: https://claude-blog.md/skills/blog-localize Cultural Localization: DACH, FR, ES, JA | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Cultural Localization // SKILL CULTURAL LOCALIZATION DEEP ADAPTATION Use this skill in the hosted vault Translation gets the words right. Localization gets the content right. A US-centric blog post citing dollar amounts, baseball metaphors, and Thanksgiving examples does not work in Germany or Japan even when accurately translated. blog-localize handles the deep-adaptation layer: idioms swapped for target-culture equivalents, examples replaced with locally familiar references, currency and units converted, cultural taboos avoided. Four built-in profiles (DACH, Francophone, Hispanic, Japanese) plus YAML-defined custom profiles. CONTRIBUTED BY CHRIS MUELLER · PRO HUB CHALLENGE · original repo $ /blog localize <file> --locale de-DE COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS FOUR LAYERS OF ADAPTATION. Localization is the layer above translation. Translation handles the words. Localization handles the cultural surface: idioms, examples, references, currency, units, formality. Without this layer, translated content reads accurately but feels foreign, and conversion rates suffer. blog-localize ships with four cultural profiles tuned by region. Each profile is a YAML file defining idiom swaps, example substitutions, currency rules, and formality registers. Profiles handle the work the translator alone cannot: a German reader does not relate to Thanksgiving, and translating it literally as Erntedankfest does not fix the cultural disconnect. 01 IDIOM SWAPS US idioms ('home run', 'ballpark figure', 'monday morning quarterback') swapped for target-culture equivalents that land naturally. 02 EXAMPLE LOCALIZATION US-centric examples (Thanksgiving, Super Bowl, July 4th) replaced with locally familiar references (Oktoberfest for DACH, Bastille Day for FR, Golden Week for JA). 03 CURRENCY + UNITS Dollar amounts converted to local currency at current rates. Imperial units converted to metric. Fahrenheit to Celsius. Mile to kilometer. Optional flags to preserve original. 04 FORMALITY REGISTERS Japanese profile honors formality registers (casual, polite, honorific). German profile honors Sie/du distinction. French profile honors tu/vous distinction. Per-audience configurable. // USAGE HOW TO RUN LOCALIZE Command What it does /blog localize <file> --locale <code> Adapts a translated file to the target culture using the matching profile Built-in profiles Profile Locale codes What it adapts DACH de-DE, de-AT, de-CH Idioms, Bundesliga and Oktoberfest references, EUR/CHF currency, metric units, Sie/du formality, DSGVO legal references Francophone fr-FR, fr-CA, fr-AFR Idioms, Bastille Day and Tour de France references, EUR/CAD currency, metric units, tu/vous formality, RGPD legal references Hispanic es-ES, es-MX, es-AR Idioms, La Liga and regional holiday references, EUR/MXN/ARS currency, metric units, tu/usted formality, regional brand swaps Japanese ja-JP Idioms, Golden Week and seasonal references, JPY currency, metric units, casual/polite/honorific registers, APPI legal references Custom profiles Define your own locale by adding a YAML config under skills/blog-localize/references/profiles/<your-locale>.yaml. The profile becomes available via the --locale flag immediately after the file is written. Minimal structure: locale: pt-BR name: Brazilian Portuguese glossary: - source: "machine learning" target: "aprendizado de maquina" idiom_swaps: - source: "ballpark figure" target: "ordem de grandeza" example_substitutions: - source: "Thanksgiving" target: "Carnaval" formality_register: default: informal b2b: formal taboos: - "politically sensitive topic" Optional flags --preserve-original keeps the original currency or unit in parentheses next to the converted value --units-source keeps imperial units instead of converting to metric --register <casual|polite|honorific> overrides the profile default for Japanese --output <path> writes to a new file instead of overwriting the input // FAQ QUESTIONS ABOUT LOCALIZE How is localization different from translation? - Translation handles language: the words change. Localization handles culture: the content changes to land naturally with a local audience. A US blog post translated to German is still about Thanksgiving, the Super Bowl, and dollar amounts. A localized version adapts those references to Oktoberfest, Bundesliga, and euros. Translation is necessary but not sufficient for global audiences. Which cultural profiles ship by default? + DACH (German-Austrian-Swiss with regional variants), Francophone (Standard French, Quebec French, African French), Hispanic (Spain Spanish, Mexican Spanish, Argentinian Spanish), and Japanese (with three formality registers: casual, polite, honorific). Each profile is a YAML config defining idiom swaps, example substitutions, formatting rules, and cultural taboos to avoid. Can I add a custom profile? + Yes. Create a YAML file at skills/blog-localize/references/profiles/<your-locale>.yaml. Define glossary (preferred translations for industry terms), idiom_swaps (source-to-target idiom mappings), example_substitutions (US-centric examples to target-culture replacements), formality_register (formal/informal language rules), and taboos (topics or phrases to avoid). The profile becomes available via --locale flag immediately. Does it handle currency conversion? + Yes. Dollar amounts are converted to local currency at the current exchange rate, with the original preserved in parentheses optionally. Use --preserve-original to keep both. Imperial units convert to metric (mile to km, mile to km, Fahrenheit to Celsius, pound to kg). Add --units-source to keep the original units. Who contributed this skill? + Chris Mueller via the AI Marketing Hub Pro Hub Challenge. blog-localize is one of four skills in the multilingual suite (write, translate, localize, locale-audit) plus the blog-translator agent. The suite scored 85/100 Proficient. Original at github.com/Chriss54/claude-blog-multilingual. // RELATED SKILLS EXPLORE MORE SEO TRANSLATION Translate handles language; localize handles culture. Pair for fully native-feeling international content. SEO TRANSLATION → MULTILINGUAL ORCHESTRATOR /blog multilingual runs translate then localize then locale-audit in sequence. Use orchestrator for end-to-end workflows. MULTILINGUAL ORCHESTRATOR → MULTILINGUAL QA After localizing, locale-audit verifies cultural adaptation completeness alongside hreflang and meta parity. MULTILINGUAL QA → VIEW ALL 30 SKILLS → LOCALIZE FOR EVERY MARKET IN ONE COMMAND. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Multilingual Blog QA: Completeness + Hreflang | Claude Blog URL: https://claude-blog.md/skills/blog-locale-audit Multilingual Blog QA: Completeness + Hreflang | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Multilingual Audit // SKILL MULTILINGUAL AUDIT QA ACROSS ALL LANGUAGES Use this skill in the hosted vault Shipping a blog post in 4 languages means maintaining 4 versions. Translations drift from the source over time. Hreflang tags break when one language gets republished. Meta descriptions go out of length on one locale and not others. Schema blocks lose a translated field after a refactor. blog-locale-audit catches all of that. Point it at your multilingual content directory and get a structured report of completeness, hreflang correctness, meta parity, and freshness across every language variant. CONTRIBUTED BY CHRIS MUELLER · PRO HUB CHALLENGE · original repo $ /blog locale-audit ./content/ COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS 4 AUDIT DIMENSIONS. Multilingual blogs decay quietly. A source post gets a freshness update, but only English ships. A schema block loses its inLanguage field after a refactor. A German meta description balloons past 160 characters because German compound words run longer than English. None of those failures throw errors. They quietly degrade international SEO over months. blog-locale-audit scans your multilingual content directory and produces a structured report across four dimensions: completeness, hreflang correctness, meta parity, and freshness. Run it standalone for ongoing QA, or wire it into CI with --strict for pre-deploy guards. 01 COMPLETENESS CHECK Lists source posts that lack translations in your target locales. Identifies orphan translations where the source post no longer exists. 02 HREFLANG VALIDATION Bidirectional hreflang check. Every language variant must reference every other variant plus x-default. Catches missing and broken hreflang relationships. 03 META PARITY AUDIT Title and description length per locale. Some languages run longer than English; checks that target-locale meta stays within Google's truncation limits. 04 FRESHNESS TRACKING Source post updated last week, but the German translation is from 3 months ago? Locale-audit flags every language variant that has drifted from the source date. // USAGE HOW TO RUN LOCALE-AUDIT Commands Command What it does /blog locale-audit <directory> Scans the directory for multilingual content and outputs a structured report across completeness, hreflang, meta parity, and freshness Output format Locale-audit emits a structured report with severity tiers per finding type (critical, high, medium, warning). Critical findings include missing hreflang return tags and absent x-default. High findings include missing translations and out-of-range meta. Medium covers slug and tag localization. Warning surfaces drifted freshness inside the configurable window. Default output is markdown. Pass --json for a machine-readable report you can pipe into CI dashboards or downstream tooling. Pass --html to also write locale-audit-report.html alongside the markdown. CI integration Pass --strict to exit non-zero on critical findings. Suitable for pre-deploy guards in GitHub Actions, GitLab CI, or any pipeline that blocks on shell exit codes. Combine with --json to publish the structured report as a CI artifact for review. /blog locale-audit ./content/ --strict --json > locale-audit.json The skill never edits your content. It only reads and reports. To fix issues, hand off to the specialist skills: /blog translate fills missing translations and refreshes stale ones, /blog localize deepens cultural adaptation, and /blog multilingual regenerates hreflang assets from scratch. // FAQ QUESTIONS ABOUT LOCALE-AUDIT What does completeness check? - Two directions. First, every source post should have translations in your configured target locales (if you publish in en, es, fr, de, audit flags source posts missing one or more). Second, every translation should have a source (orphan translations where the original post got deleted are flagged as cleanup candidates). The completeness report is a single source of truth for what is shipped where. How is hreflang validation done? + blog-locale-audit verifies hreflang relationships are bidirectional. If the English post links to the German variant via hreflang, the German variant must link back to English. Missing return tags break international targeting. x-default tag presence is also checked (recommended for sites with a primary language fallback). Output flags every broken or missing relationship. What is meta parity? + Meta description length varies by language. A 145-character English description might run 175 characters when translated to German (longer compound words). Meta parity audit checks every language variant's title and meta description against the 120-160 character target. Out-of-range variants are flagged for retranslation. How does freshness tracking work? + Source post updated last week, but the German translation is from 3 months ago? Locale-audit reads dateModified or article:modified_time meta from every variant and flags any translation that has not been updated since the source post's last update. Configurable freshness window (default: flag translations older than 30 days behind source). Who contributed this skill? + Chris Mueller via the AI Marketing Hub Pro Hub Challenge. blog-locale-audit is one of four skills in the multilingual suite (write, translate, localize, locale-audit) plus the blog-translator agent. The suite scored 85/100 Proficient. Original at github.com/Chriss54/claude-blog-multilingual. Chris also contributed seo-hreflang cultural profile enhancements to claude-seo v1.9.0. // RELATED SKILLS EXPLORE MORE MULTILINGUAL ORCHESTRATOR Use the orchestrator end-to-end; use locale-audit standalone for ongoing QA on existing multilingual content directories. MULTILINGUAL ORCHESTRATOR → SEO TRANSLATION Locale-audit detects translations that need refresh; pair with /blog translate to update drifted variants. SEO TRANSLATION → CULTURAL LOCALIZATION Locale-audit checks structural QA; localize handles cultural adaptation. Run both during multilingual publishing. CULTURAL LOCALIZATION → VIEW ALL SKILLS → AUDIT YOUR MULTILINGUAL BLOG IN 30 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Blog Brand: BRAND.md + VOICE.md Context | Claude Blog URL: https://claude-blog.md/skills/blog-brand Blog Brand: BRAND.md + VOICE.md Context | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Blog Brand // SKILL BLOG BRAND ONE EDITORIAL SOURCE OF TRUTH Use this skill in the hosted vault Stop re-explaining your brand to every command. /blog brand generates two project-root files, BRAND.md and VOICE.md, that every Claude Blog sub-skill auto-loads before it writes or reviews. Audience, positioning, editorial do/don't rules, taboo phrases, competitor differentiation, and voice, captured once and carried everywhere. Free with Claude Blog. $ /blog brand init COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // WHY IT EXISTS CONTEXT THAT SURVIVES SESSIONS. Persona JSON is loaded by some skills and not others. Topic-cluster context lives inside cluster vaults. Competitor positioning lives nowhere. Without a canonical source, each blog command re-derives "what is the brand" from whatever scraps of context it happens to have. BRAND.md and VOICE.md fix that. They are the editorial equivalent of a persistent PRODUCT.md or DESIGN.md: one canonical source, written to the project root and loaded by the blog orchestrator at the start of every command. Run it once and every sub-skill, from write to audit, works from the same brief. 1 BRAND.md Who the audience is, what the brand stands for, the editorial do/don't rules, taboo phrases to never use, and how you differ from competitors. The strategic brief. 2 VOICE.md How the brand sounds, structurally and lexically. Your existing persona JSON re-expressed as readable prose a writer or reviewer can apply directly. ∞ AUTO-LOADED When present, both files are loaded by the orchestrator before every command. No flag, no re-pasting. Write, rewrite, analyze, and audit all share the same context. ✓ BACKWARD COMPATIBLE No BRAND.md or VOICE.md? Behavior is unchanged from earlier versions. The files are purely additive. Nothing breaks if you never run the command. // USAGE HOW TO RUN IT Commands Command What it does /blog brand init Runs an interactive interview, then writes BRAND.md and VOICE.md to the project root /blog brand show Displays the current contents of both files, or reports them missing /blog brand update Re-runs the interview with current values pre-filled as defaults The Interview The init flow is a short, step-by-step interview. It asks one section at a time and waits for your answer before moving on, covering audience (role, expertise level, the problems readers are solving, common misconceptions), positioning and differentiation, editorial do/don't rules and taboo phrases, and brand voice. If a blog-persona JSON already exists, voice answers are pre-filled from it so you do not repeat yourself. How Sub-Skills Use It Once written, the files propagate automatically: /blog write and /blog rewrite apply the voice and respect the taboo phrases while drafting. /blog analyze and /blog audit score against the brand's stated rules, not generic defaults. /blog brief and /blog strategy anchor positioning and audience to the same source. Pairs With blog-persona blog-persona manages the structured persona JSON; blog-brand re-expresses that persona as readable VOICE.md prose and adds the strategic BRAND.md layer on top. Use persona for machine-readable voice profiles and brand for the human-readable canonical brief. // FAQ QUESTIONS ABOUT BLOG BRAND What does /blog brand do? - It generates two project-root files: BRAND.md (who the audience is, what the brand stands for, editorial do/don't rules, taboo phrases, competitor differentiation) and VOICE.md (how the brand sounds, expressed as readable prose). When these files exist, every Claude Blog sub-skill auto-loads them before writing or reviewing, so each command works from one canonical source of brand context instead of re-deriving it. What are the blog brand commands? + Three: /blog brand init runs an interactive interview and writes BRAND.md and VOICE.md to the project root; /blog brand show displays the current contents (or reports them missing); /blog brand update re-runs the interview with current values as defaults. How does blog-brand relate to blog-persona? + They pair. blog-persona manages the structured persona JSON; blog-brand re-expresses that persona as readable VOICE.md prose and adds BRAND.md for audience, positioning, and editorial rules. If a persona JSON already exists, the brand interview pre-fills voice answers from it. Is blog-brand backward compatible? + Yes. When neither BRAND.md nor VOICE.md exists, behavior is unchanged from earlier versions. The files are purely additive context; nothing breaks if you never run /blog brand. Where are BRAND.md and VOICE.md written? + Both are written to the project root so the blog orchestrator loads them at the start of every command. They are the editorial equivalent of a persistent PRODUCT.md or DESIGN.md: context that survives across sessions and propagates to every sub-skill. // RELATED SKILLS EXPLORE MORE WRITING PERSONAS Manage structured persona JSON and voice profiles per brand. blog-brand re-expresses persona as VOICE.md prose. PERSONA MANAGER → BLOG STRATEGY Niche positioning, topic ideation, and content pillar architecture, anchored to the same BRAND.md audience and positioning. STRATEGY PLANNER → BLOG WRITER Drafts full articles that apply the voice and respect the taboo phrases captured in your brand context automatically. CONTENT WRITER → VIEW ALL 30 SKILLS → CAPTURE YOUR BRAND ONCE. USE IT EVERYWHERE. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Blog Chart: Inline SVG Data Visualization | Claude Blog URL: https://claude-blog.md/skills/blog-chart Blog Chart: Inline SVG Data Visualization | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Blog Chart // SKILL BLOG CHART DATA THAT RENDERS ITSELF Use this skill in the hosted vault When a post has chart-worthy data, Claude Blog draws it. /blog chart generates dark-mode-compatible inline SVG, seven chart types selected by the shape of your data, with accessible markup and source attribution baked in. No charting library, no external request. It runs automatically inside the writing pipeline. Free with Claude Blog. $ /blog chart COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // HOW IT WORKS INLINE SVG. NO LIBRARY. blog-chart is built into the writing pipeline. blog-write and blog-rewrite invoke it automatically whenever they identify chart-worthy data: three or more comparable metrics, a trend over time, or before/after numbers. You do not call it by hand. The chart appears in context as the post is built. Output is plain inline SVG. No charting library loads, no external request fires, so the page stays light and passes strict content security policies. Backgrounds are transparent, so the chart inherits the page theme and stays dark-mode-compatible without a second render. The platform is detected automatically, raw HTML or JSX/MDX, and the markup is shaped to match. 7 CHART TYPES Horizontal bar, grouped bar, donut, line, lollipop, area, and radar. The type is chosen by the shape of the data, and never repeated within a single post. ☽ DARK-MODE NATIVE Transparent backgrounds mean the chart inherits the page theme. No second light/dark render, no mismatched panel behind your data. A ACCESSIBLE SVG Every chart carries role=img and a descriptive aria-label, plus visible source attribution under the figure. Readable by screen readers and by AI crawlers. </> AUTO PLATFORM Detects HTML versus JSX/MDX and emits markup that drops straight in. The same chart request works across static sites and framework blogs. // CHART SELECTION THE RIGHT TYPE FOR THE DATA Chart type is not a style choice; it follows the data pattern. blog-chart maps each pattern to the type that reads cleanest, and enforces diversity so no single post leans on one shape. Data Pattern Chart Type Before / after comparison Grouped bar Ranked factors or correlations Lollipop Parts of a whole / market share Donut Trend over time Line Percentage improvement Horizontal bar Distribution / range Area Multi-dimensional scoring Radar Chart Request Format Internally, the writer or researcher passes a structured request that blog-chart turns into SVG: Chart Request: - Type: horizontal bar - Title: "AI Citation Sources by Platform" - Data: ChatGPT 43.8%, Perplexity 6.6%, ... - Source: Ahrefs, December 2025 - Platform: mdx (or html) Non-Negotiable Styling Transparent background so the chart works on any theme. Type diversity enforced; a chart type is never repeated inside one post. Source attribution rendered with every chart so the data is verifiable. Accessible markup with role=img and aria-label on every figure. // FAQ QUESTIONS ABOUT BLOG CHART What does blog-chart do? - It generates dark-mode-compatible inline SVG charts for blog posts. It supports seven chart types (horizontal bar, grouped bar, donut, line, lollipop, area, and radar), detects the target platform automatically (raw HTML versus JSX/MDX), and emits accessible markup with role=img and aria-label, source attribution, and transparent backgrounds that work on any theme. Is blog-chart a command I run directly? + It runs primarily inside the writing pipeline. blog-write and blog-rewrite invoke it automatically whenever they identify chart-worthy data, such as three or more comparable metrics, a trend over time, or before/after numbers. You do not need to call it manually; it produces the right visualization in context as the post is built. What chart types are supported? + Seven: horizontal bar (percentage improvements), grouped bar (before/after comparisons), donut (parts of a whole or market share), line (trends over time), lollipop (ranked factors or correlations), area (distributions or ranges), and radar (multi-dimensional scoring). Chart type is selected by the shape of the data, and diversity is enforced so a type is never repeated within a single post. Are the charts accessible? + Yes. Every chart ships with accessible SVG markup including role=img and a descriptive aria-label, plus visible source attribution. Backgrounds are transparent so the chart inherits the page theme, which keeps it dark-mode-compatible without a second render. Do the charts require any external library? + No. Charts are emitted as plain inline SVG with no JavaScript charting library and no external request. They render anywhere the post does (static HTML, MDX, or a framework), keep the page lightweight, and pass strict content security policies. // RELATED SKILLS EXPLORE MORE BLOG WRITER Drafts full articles and calls blog-chart automatically when the draft contains chart-worthy data points. CONTENT WRITER → CONTENT REWRITER Optimizes existing posts and adds inline SVG charts where comparable metrics or trends appear in the source. CONTENT OPTIMIZER → SCHEMA MARKUP Generates JSON-LD structured data. Pairs with charts to make data both human-readable and machine-parseable. SCHEMA GENERATOR → VIEW ALL 30 SKILLS → TURN YOUR DATA INTO CLEAN SVG. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Blog Discourse: API-Free 30-Day Research | Claude Blog URL: https://claude-blog.md/skills/blog-discourse Blog Discourse: API-Free 30-Day Research | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Skills/Blog Discourse // SKILL BLOG DISCOURSE WHAT PEOPLE ARE ACTUALLY SAYING Use this skill in the hosted vault Authority sources tell you what is true. Discourse tells you what is fresh. /blog discourse researches the last 30 days across Reddit, X, YouTube, Hacker News, dev.to, and Medium, then writes a structured DISCOURSE.md brief your writer can consume. No API keys. Just WebSearch with platform-targeted operators. Free with Claude Blog. METHODOLOGY ADAPTED FROM last30days-skill · Matt Van Horn, MIT · github.com/mvanhorn/last30days-skill $ /blog discourse <topic> COPY REQUIRES CLAUDE BLOG INSTALLED IN CLAUDE CODE // WHY IT EXISTS THE RECENCY LENS RESEARCH LACKS. blog-researcher, the authority-first research agent, finds statistics, verifies sources, and grounds claims in citable references. That is the right lens for accuracy, but it misses what is happening this month: the complaints, the workarounds, the questions practitioners keep asking, the language real customers use. blog-discourse is the recency-and-engagement lens. It asks one question: in the last 30 days, what are practitioners and customers actually saying about this topic on the public web? It uses WebSearch with platform-targeted site operators plus recency filters, so it needs no platform APIs and no keys. The output is DISCOURSE.md plus JSON the writer can consume directly. 30 DAY WINDOW Recency filters keep the brief fresh. Widen to 90 days with --days when a slower-moving topic needs a broader window. 6+ PLATFORMS Reddit, X/Twitter, YouTube, Hacker News, dev.to, Medium, and other public discourse, reached with site operators, not paid APIs. ♬ DISCOURSE.md A structured brief plus JSON: themes, recurring questions, objections, and the language customers actually use, ready for the writer. → FEEDS THE WRITER --feed-into brief, write, or strategy runs the research and hands DISCOURSE.md straight to the next command, auto-loaded. // USAGE HOW TO RUN IT Commands Command What it does /blog discourse <topic> Produces a discourse brief at project-root DISCOURSE.md /blog discourse <topic> --days 90 Widens the freshness window from 30 to 90 days /blog discourse <topic> --feed-into brief Runs the brief, then invokes /blog brief with DISCOURSE.md loaded /blog discourse <topic> --feed-into write Runs the brief, then invokes /blog write /blog discourse <topic> --feed-into strategy Runs the brief, then invokes /blog strategy /blog discourse <topic> --input results.json Skips search; builds the brief from a pre-gathered results file Topic Pre-Flight Before any search, discourse runs keyword-trap checks so it does not waste WebSearch calls on a query that will only return noise. Topics that match a trap class (demographic shopping, a numeric trap, an overly literal phrase, or a generic single noun) are either reframed automatically or paused for a one-line clarifying question. Topic Decomposition For named-entity topics, the query is decomposed into discrete searches: the primary entity (official statements and vendor pages), the counter-perspective (critics, competitors, contrarians), and practitioner discourse (subreddits, forums, dev.to, Medium). The brief is assembled from all three so it is balanced, not just the loudest take. Attribution blog-discourse adapts the methodology of last30days-skill by Matt Van Horn (MIT licensed). The upstream uses platform APIs; this sub-skill re-implements the approach with WebSearch and site operators so it requires no API keys, while preserving attribution to the original work. // FAQ QUESTIONS ABOUT BLOG DISCOURSE What does /blog discourse do? - It researches what practitioners and customers are actually saying about a topic in the last 30 days across Reddit, X/Twitter, YouTube, Hacker News, dev.to, Medium, and other public platforms. It is the recency-and-engagement lens that authority-first research lacks, and it writes a structured brief, DISCOURSE.md, plus JSON output the writer can consume. Does blog-discourse need API keys? + No. It is API-free. It uses WebSearch with platform-targeted site operators and recency filters instead of paid platform APIs, so there are no keys to configure and no per-call cost. It can also build a brief from a pre-gathered results file with --input. How is it different from blog-researcher? + blog-researcher is authority-first: it finds statistics and verifies sources. blog-discourse is recency-and-engagement-first: it captures what people are saying right now on the public web. They complement each other; discourse adds voice-of-customer and trend signal that authority sources miss. Can discourse research feed straight into writing? + Yes. Use --feed-into brief, write, or strategy to run the discourse brief and then immediately invoke that command with DISCOURSE.md auto-loaded. You can also widen the freshness window with --days 90 when 30 days is too narrow. Where did the methodology come from? + It adapts the methodology of the last30days-skill by Matt Van Horn (MIT licensed). The upstream uses platform APIs; blog-discourse re-implements the approach with WebSearch and site operators so it needs no API keys, while preserving attribution. // RELATED SKILLS EXPLORE MORE CONTENT BRIEF Detailed briefs with competitive analysis. --feed-into brief hands DISCOURSE.md straight to it for a recency-grounded outline. BRIEF BUILDER → BLOG STRATEGY Positioning and content planning. Discourse signal shows which angles are live before you commit a quarter of content. STRATEGY PLANNER → FACT CHECK Discourse surfaces claims people repeat; factcheck verifies them against sources before they reach a published post. FACT CHECKER → VIEW ALL 30 SKILLS → HEAR THE CONVERSATION BEFORE YOU WRITE. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL SKILLS > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG --- ## Claude Watermarks + SynthID: Marketer's Guide | Claude Blog URL: https://claude-blog.md/blog/claude-watermark-synthid-guide Claude Watermarks + SynthID: Marketer's Guide | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/Claude Watermarks and SynthID Marketer's Guide // PUBLISHED UNDER CLAUDE-BLOG QUALITY GATES · Researched from primary sources on August 20, 2026: Anthropic, Google DeepMind, EU AI Act texts, Nature, and current platform policies. See how the gates work → // DEEP DIVE CLAUDE NOW WATERMARKS ITS TEXT WHAT SYNTHID MEANS FOR MARKETERS DANIEL AGRICI // PUBLISHED AUG 20, 2026 // 9 MIN READ // DEEP DIVE AI WATERMARKING SYNTHID MARKETING // Audio summary Narrated by Ava · 1:11 Your browser does not support audio playback. Download the summary. Read the transcript In August 2026, Anthropic started adding an invisible watermark to the text that Claude generates. It is based on Google DeepMind's SynthID, and it works by nudging the model's word choices into a pattern that only shows up if you hold the secret key. You cannot see it, and it does not change how the writing reads. Why now? The European Union's AI Act made machine readable marking of AI content mandatory from August 2026, with fines that reach into the millions, and nearly every major AI company has signed on. So what does this mean if you use AI for content? Less than the panic suggests. Google still ranks pages on quality, not on whether a machine helped write them. The watermark says nothing about who owns the work, and there is no public detector yet. It survives copy and paste and light edits, but a genuine rewrite in your own voice removes it. That real human editing is exactly what the law rewards and what readers trust. The takeaway is simple. Keep AI in the drafting seat, keep a human in the editor's chair, and let quality carry the page. Provenance is becoming infrastructure. Quality is still the product. Key Takeaways • AI watermarking is now default infrastructure: Google marks output across its products, Anthropic marks Claude's text, and detection surfaces are coming to Search and Chrome. • Watermarks change disclosure and platform labeling, not Google's quality rules. Content is still judged on quality and E-E-A-T, not production method. • Workflows with real human editing behave differently from publish-raw workflows: heavy editing removes the signal, and the EU AI Act exempts human-reviewed text. Anthropic has started adding an invisible, machine-readable watermark to text generated by Claude, using technology adapted from Google DeepMind's SynthID-Text. If AI is part of your content workflow, this matters. Major AI providers are making generated text easier to identify, the EU AI Act has required machine-readable marking since August 2, 2026, and platforms are already labeling AI-generated media. Researched from primary sources on August 20, 2026, this guide breaks down what changed, how the watermark works, where it falls short, and what marketing teams should do next. To be clear, this doesn't mean AI-assisted content will be penalized. Watermarking is about tracing where content came from and whether its use is disclosed, not judging its quality. But there will be a clearer distinction between publishing raw AI output and using AI as part of a genuinely human-led editing process. That difference already matters technically and legally, and soon it may be much easier to measure. What Anthropic actually shipped In mid-August 2026, Anthropic announced that Claude's text output now carries a machine-readable watermark. The change applies worldwide, not just in the EU. According to Anthropic's help center, Claude models launched on or after August 2, 2026 support marking at launch, while models launched before that date fall under a legal transition period while marking is added. No user opt-out is documented. The mechanism is an adaptation of Google DeepMind's SynthID-Text. When Claude chooses among several comparably likely next words, the watermark changes only the source of randomness used for that choice, steering it with a key plus the preceding words. The result reads identically to unwatermarked text, is invisible to readers, and is detectable only with the key. Anthropic reports no impact on quality, speed, or price. Equally important is what a detection result does not prove. Anthropic is explicit that detection cannot confirm a text was human-written, cannot identify output from other AI systems, cannot distinguish text Claude wrote from text Claude heavily edited, says nothing about ownership or authorship, and identifies nothing about individual users. It answers exactly one question: does this passage carry Claude's statistical signature? SOURCE: GOOGLE DEEPMIND SYNTHID PAGE, CAPTURED AUG 20, 2026 How SynthID-Text works, in plain language SynthID is Google DeepMind's watermarking family. It covers images, audio, video, and text, and already runs across Google products: text in the Gemini app and on the web, audio from Lyria, and outputs in NotebookLM. For text, SynthID works by adjusting the probability scores of candidate tokens as the model generates. Google DeepMind explains SynthID (official video) Skip the math. A language model generates text one word at a time. At many points in a sentence, multiple continuations are nearly equally plausible. For instance, if asked to complete a sentence about a favorite tropical fruit, options like mango, banana, or papaya might all be reasonable. An unwatermarked model selects among these options using standard randomness. A watermarked model also makes a selection, but its randomness is derived from a keyed function of the preceding words. No single word choice gives the watermark away. But across hundreds of words, these subtly biased choices accumulate into a statistical signature. This signature can be detected by someone who holds the key, while remaining invisible to everyone else. This is not a lab prototype. The method was peer reviewed in Nature in October 2024 as "Scalable watermarking for identifying large language model outputs" (Nature 634, 818-823). The paper includes a live experiment on nearly 20 million Gemini responses, which found no significant difference in user feedback between watermarked and unwatermarked output. DeepMind also documents the limits. Thorough rewriting or translation defeats the watermark, and factual prompts with little word-choice variation carry a weak signal. A question about the capital of France has one good answer, so there are no comparably likely words for the watermark to work with. Watermarking is already everywhere Watermarking sounds like a compliance footnote until you look at the volume Google reports. When the SynthID Detector was announced in May 2025, Google said over 10 billion pieces of content had already been watermarked. By November 2025, the figure was over 20 billion pieces. At Google I/O in May 2026, the keynote cited over one hundred billion images and videos, plus sixty thousand years of audio. Note that the units shift between announcements, so treat these as milestones, not one continuous series. SynthID scale milestones as reported by Google. Units differ across reports: pieces of content in 2025, images and videos plus audio in 2026. Why now: the EU AI Act made it mandatory Article 50(2) of the EU AI Act requires providers of generative AI systems to ensure their outputs are marked in a machine-readable format and detectable as artificially generated, applicable from August 2, 2026. That is the exact date Anthropic uses to split models that mark at launch from models in transition. Non-compliance is expensive: penalties run up to EUR 15 million or 3% of total worldwide annual turnover, whichever is higher. The Act also contains the clause that matters most for content teams: Article 50 includes an exemption for AI-generated text that has undergone human review and editorial control. European regulators drew a line between raw machine output and edited work that a person takes responsibility for. That distinction should shape your workflow more than the watermark itself. Industry has moved with the regulation rather than against it. By July 31, 2026, about 190 organizations had signed the EU Code of Practice on Transparency of AI-Generated Content, including Anthropic, Google, Meta, Microsoft, and OpenAI. SOURCE: DEEPMIND.GOOGLE/MODELS/SYNTHID, CAPTURED AUG 20, 2026 SynthID vs C2PA: two different kinds of provenance Claude actually ships two provenance systems, and they behave differently. Text gets the statistical watermark described above. Files Claude produces, such as .png, .jpg, and .svg exports, instead get C2PA provenance metadata: cryptographically signed, tamper-evident, and attached to the file. Because it sits in metadata rather than in the content itself, it can be stripped. C2PA is the Coalition for Content Provenance and Authenticity, the open standard behind Content Credentials. Its steering committee includes Adobe, BBC, Google, Intel, Microsoft, OpenAI, Sony, and Truepic, and OpenAI attaches C2PA credentials to DALL-E 3 images. The two approaches compare like this. Property SynthID text watermark C2PA Content Credentials What it is A statistical pattern in the word choices themselves Cryptographically signed provenance metadata Where it lives Inside the content In the file, alongside the content Survives copy-paste? Yes, it travels with the text Only if the file and its metadata travel together Survives screenshot? In substance, yes: retyping the same words preserves the pattern No, a screenshot creates a new file without credentials Survives metadata stripping? Yes, there is no metadata to strip No, stripping the metadata removes it Who uses it Google products; Claude text output Adobe, BBC, Google, Intel, Microsoft, OpenAI, Sony, Truepic; Claude file outputs What this means for marketing teams Marketer's take0:17 Google Search does not care how you wrote it Google's published position is stable: appropriate use of AI is not against its guidelines, and content is judged on quality and E-E-A-T regardless of how it was produced. The red line is scaled content abuse, meaning mass-produced low-value pages created primarily to manipulate rankings. Nothing in the watermark rollout changes that stance. If your AI-assisted pages are useful, original, and edited by someone accountable, the production method is not the ranking variable. Quality is. Platform labels target media first Labeling is already routine for images and video. Meta applies AI labels on Facebook, Instagram, and Threads using industry-standard indicators from companies like Google, OpenAI, Microsoft, and Adobe. TikTok requires creators to label realistic AI-generated content and automatically labels media carrying C2PA Content Credentials. YouTube requires disclosure of realistic synthetic content. Automatic labeling of watermarked text has not been deployed on any major platform, but the media precedent shows where the infrastructure is heading. Consumers punish visible AI, not hidden provenance AI adoption in marketing is close to universal. HubSpot's 2026 State of Marketing report, based on a survey of more than 1,500 marketers, found that 86.4% of marketing teams use AI in at least a few areas. For content creation specifically, 42.5% use it extensively and another 38% use it occasionally. Consumers are much less positive about it. A Klaviyo and Datalily survey of 8,000 people across eight countries, conducted in December 2025 and reported by eMarketer, found that visibly AI-generated marketing increases trust for just 7% of consumers. For 31%, it reduces trust. The key word is "visibly." The watermark itself can't be seen; what people react to is content that feels obviously machine-made. The answer isn't to avoid the watermark, it's to edit the content properly. Consumer response to visibly AI-generated marketing. Klaviyo and Datalily survey of 8,000 consumers in 8 countries, via eMarketer. The limits: what a watermark can and cannot tell you The limits, fast0:17 The watermark is robust to normal reuse but not to real editing. Based on Anthropic's documentation and TechCrunch's reporting on the rollout, the split is simple. What tends to survive: copy-paste, because the pattern travels with the text, and light editing, which usually leaves the watermark detectable. Longer passages give higher detection confidence. What weakens or removes it: a complete rewrite removes the watermark, heavy human editing leaves little signal, short samples are unreliable to test, factual low-entropy passages carry a sparse signal, and code carries little signal because the watermark attaches mainly to comments. Independent research points the same direction. ETH Zurich's SRI Lab probed SynthID-Text in December 2024 and found that spoofing, meaning forging the watermark onto text the model never wrote, is hard, with about 4% success at the base query budget. Scrubbing is the comparative weakness: paraphrasing combined with watermark-stealing techniques achieved over 90% removal success. DeepMind itself notes that thorough rewriting or translation defeats the watermark. So treat the watermark as transparency infrastructure, not forensic proof. A positive detection tells you Claude was involved somewhere in a text's history. It cannot tell you whether Claude drafted it or merely edited it, cannot name a user, and its absence proves nothing about human authorship. Where this goes next The long game0:14 At I/O 2026, Google said SynthID verification is expanding to Search and Chrome, and that OpenAI, Kakao, and ElevenLabs are adopting SynthID. Anthropic says it is working to enable users and third parties to detect Claude's watermarks, with technical documentation to follow. Detection is private today. Plan for a world where it is not. The durable playbook follows from everything above, and it is the same playbook that already wins on quality: Use AI as the drafting layer. Let models produce structure and first drafts, then verify every claim before publishing. A source verification pass exists for exactly this step. Keep a human as the accountable author and editor. Real editorial control is what the EU AI Act's human-review exemption for text recognizes, and it is what separates your pages from raw output. Disclose where platforms require it. Realistic synthetic media already carries labeling obligations on YouTube, TikTok, and Meta, whoever made it. Compete on quality, because that is the only Google-side variable. Dual optimization for search and AI answers, AI citation readiness, and a gated content pipeline matter more than any watermark ever will. Final verdict Don't fear the watermark, and don't waste time trying to bypass it. Anthropic's system doesn't change what makes content rank. It mainly reveals workflows that publish raw AI output at scale, an approach that was already falling short on quality and consumer trust. Use AI to create the first draft, but keep a responsible human in charge of editing, fact-checking, and final approval. Label synthetic media when platforms require it, then focus on publishing useful, accurate, well-edited content. Provenance is becoming part of the infrastructure, but quality is still what matters. // AUTHOR About the Author: Daniel Agrici Daniel Agrici is an AI Automation Specialist. He builds open-source systems for AI content workflows, code review, SEO, and knowledge management. He writes about evidence-led automation and production guardrails. Follow his work on GitHub and YouTube, join the AI Marketing Hub, or contact Daniel through LinkedIn. // FAQ FREQUENTLY ASKED QUESTIONS Does Claude watermark everything I write with it now? - Claude models launched on or after August 2, 2026 apply the machine-readable watermark at launch, and models launched before that date are in a legal transition period while marking is added. The change applies worldwide and Anthropic documents no user opt-out. In practice, text you heavily edit or fully rewrite carries little to no detectable signal, because the watermark lives in the model's original word choices. Can readers, clients, or Google see the watermark? + No. The pattern is invisible to human readers and requires the watermarking key to detect. Public detection tools are not yet available, though Anthropic says it is working to enable users and third parties to detect the marks. Google's published stance is that content is judged on quality and E-E-A-T, not on how it was produced. Does the watermark survive editing and copy-paste? + It travels with copy-paste, and light editing usually leaves it detectable. A complete rewrite removes it. The signal is weak on short snippets, on factual passages with little word-choice variation, and in code, where it attaches mainly to comments. Longer passages give higher detection confidence. Will platforms label my AI-assisted content automatically? + For images and video, sometimes. Meta applies AI labels using industry-standard indicators, and TikTok automatically labels media carrying C2PA Content Credentials. Automatic labeling of watermarked text is not deployed on major platforms today. YouTube requires you to disclose realistic synthetic content yourself. What is the difference between SynthID and C2PA? + SynthID embeds a statistical pattern inside the content itself, in word choices for text or directly in generated media, so it travels wherever the content goes. C2PA Content Credentials are cryptographically signed metadata attached to a file: tamper-evident, but strippable because they sit outside the content. The two have different survival properties and are often used together. // RELATED KEEP READING DUAL OPTIMIZATION GUIDE Rank in Google and get cited by AI answer engines with one workflow. The strategy side of the provenance story. READ ARTICLE → AI CONTENT PIPELINE GUIDE Build a gated drafting, editing, and verification pipeline that keeps a human accountable for every page. READ ARTICLE → BLOG FACTCHECK SKILL The source verification pass this article recommends before any AI-assisted draft goes live. VIEW SKILL → STAR CLAUDE-BLOG ON GITHUB MIT-licensed and free forever. Read the source, file issues, or star the repo to follow along with future releases. VIEW REPO → RUN THIS YOURSELF IN ABOUT 10 MINUTES. $ curl -sSL https://raw.githubusercontent.com/AgriciDaniel/claude-blog/main/install.sh | bash COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## Public Questions and PAA SEO for AI Answers | Claude Blog URL: https://claude-blog.md/blog/public-questions-paa-skill Public Questions and PAA SEO for AI Answers | Claude Blog CLAUDE BLOG Skills Blog GitHub Home/Blog/Public Questions // BLOG STRATEGY Public questions are the missing bridge between keyword volume and AI answers. The mistake is treating People Also Ask questions like normal keywords. Exact PAA strings often show little or no Google Ads volume, but the seed term carries demand. The job is to use seed demand for targeting and public questions for answer surfaces. Daniel Agrici//May 28, 2026//Strategy 16 seed SERPs checked 27 unique PAA questions DataForSEO US / English Packaged 2026-05-28 What is the public-questions workflow? The public-questions workflow turns a high-volume seed keyword into a page structure made from real questions people see in search. It starts with the seed, extracts People Also Ask and related questions, clusters them by intent, and writes answer-first sections that can stand alone as useful passages. This does not replace normal keyword research. It fixes the gap between keyword tools and actual language. A seed like claude code skills had 9,900 monthly US searches in the same-day DataForSEO evidence pack. Its PAA surface included questions such as "Do skills work in Claude Code?", "What are the best skills for Claude Code?", and "How to give Claude Code a skill?" Those questions are the article sections. Seed demand supplies the target. PAA supplies the public language. Why exact PAA questions can show zero volume Long natural-language questions often do not behave like head terms in keyword tools. That does not make them useless. It means the page should rank for the seed term while answering the question variants clearly enough to win snippets, PAA clicks, and AI answer reuse. Seed term Demand signal Answer surface ai visibility tool 1,300 volume, KD 16 What is the best AI visibility tool? How do I check AI visibility? llm visibility tool 590 volume, KD 18 What is LLM visibility? How do I track LLM visibility? claude code skills 9,900 volume, KD 32 Do skills work in Claude Code? How do I give Claude Code a skill? answer engine optimization tools 260 volume, KD 5 What is AEO vs SEO? How do I optimize for AI search? How to build a PAA-led article Choose one seed. Pick a term with measurable demand and a clear intent. Pull the SERP. Capture PAA questions, related searches, AI Overview presence, and the top competing pages. Normalize questions. Remove duplicates and group by definition, comparison, how-to, best/tools, troubleshooting, and commercial intent. Write answer-first H2s. Each section starts with a direct 40-80 word answer, then adds proof and steps. Add visible sources. Cite official docs or first-hand evidence. Avoid unsupported claims about AI ranking shortcuts. Recheck visibility. After publishing, rerun SERP and AI visibility checks for the seed and the question set. What official guidance supports this? Google's guidance for AI features points back to strong foundational SEO: make content crawlable, indexable, useful, and eligible for snippets. Google also says structured data should match visible page content. That means the safe public-questions approach is not hidden schema stuffing. It is visible, answer-first content that aligns with the user's real question. For AI-specific measurement, DataForSEO documents AI Optimization endpoints for LLM responses, ChatGPT scraping, LLM mentions, and Google AI Mode results. Those are measurement inputs. They are not guarantees. They should be logged with query, date, market, and source URLs. Schema note: keep visible FAQ-style sections for users and AI passage clarity, but do not rely on new FAQ rich results as the Google win. Use BlogPosting and BreadcrumbList schema for the article, and keep the question answers visible in the body. Where this first batch uses public questions The first AI visibility rank loop uses public questions in three places: AI Visibility Tool answers the commercial tool and tracking questions. This page explains the PAA method and why a future public-questions skill makes sense. The Claude/Codex skills ecosystem article answers the education wedge around skills, plugins, hooks, and agents. Frequently asked questions What is People Also Ask SEO? People Also Ask SEO is the practice of using the questions Google surfaces around a seed query to structure clearer answers. The page still needs a real keyword target, but the PAA questions reveal how people phrase follow-ups and objections. Should every PAA question become its own page? No. Most PAA questions should become sections inside a stronger seed-targeted page. Create separate pages only when the question has distinct intent, enough demand, or a different conversion path. Can this become a new skill? Yes, but after repeated runs. The stable skill should accept a seed keyword, pull PAA and keyword evidence, cluster questions, write answer-first blocks, and output schema-ready page briefs. For now, Claude Blog and Claude SEO already cover enough of the workflow to publish the first batch. Claude Blog by Daniel Agrici. Last updated May 28, 2026. llms.txt Blog Strategy skill GitHub --- ## Claude Blog v1.9.1: 5-Gate Contract + 95/100 Audit URL: https://claude-blog.md/blog/claude-blog-v191-release Claude Blog v1.9.1: 5-Gate Contract + 95/100 Audit Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/Claude Blog v1.9.1 Release // CHANGELOG CLAUDE BLOG V1.9.1 5 GATES, 95/100, NO SLOPPY DRAFTS DANIEL AGRICI // MAY 21, 2026 // 9 MIN READ // CHANGELOG SECURITY RELEASE v1.9.1 is two releases in one announcement. v1.9.0 made the blog reviewer blocking instead of advisory, so drafts under 90 of 100 stop existing on disk before they ever reach you. v1.9.1 ran the whole project through an 8-agent cybersecurity audit and closed every HIGH and MEDIUM finding in 24 hours. Both shipped this week. Both ship together because they are the same shift: the gates do the work the human used to have to do. If you came here from the walkthrough video, the timeline is real. While I was talking through the contract, an agent was generating the Codex vs Claude sample blog on screen in another window. Ten minutes, end to end, including research, hero image, SVG charts, and a passing review against the 5 gates. That post is on the site now. Skim it first if you want to see what the contract produces before reading what the contract does. The Blog Delivery Contract: 5 Gates Before Any Draft Reaches You Until v1.9.0, the architecture had a real defect. The blog-reviewer agent existed. It scored drafts. It produced a scorecard. And then the orchestrator handed the draft to you anyway, regardless of the score. Sloppy drafts with no real hero image, broken og:image, SVGs that overflowed on mobile, missing PDFs, and md/html divergence all reached the user before any gate fired. The user ended up being the first reviewer. v1.9.0 closes that gap. The contract has 5 gates. Gate 4 is the blocking one. Gate What it does Blocks? 1. Capability Discovery Enumerates MCP servers, agents, scripts, env vars before drafting. Writes capabilities.json. No (info only) 2. Format Completeness Every draft must ship .md + .html + .pdf + a real hero image. Deterministic renderer from one source. Yes 3. Visual Verification Playwright renders the HTML at 375, 768, and 1280 px. Asserts no SVG overflow, dark-mode swap works, console clean. Yes 4. Content Review (BLOCKING) blog-reviewer scores the rendered HTML. Threshold: 90 of 100, zero P0 issues. Emits machine-readable BLOCKING: true|false. Yes (hard) 5. Asset + Link Integrity Every img src resolves, og:image points to a real file, external links return 200, JSON-LD wordCount matches body within 5%. Yes On any gate failure, the orchestrator iterates up to 3 times before escalating to the user with a diagnostic. You are no longer the first reviewer. The gates are. The Hero Image Ladder: 5 Tries Before We Block Gate 2 demands a real hero image. The old failure mode was a missing image plus a fabricated og:image URL that 404d the moment someone shared the post on LinkedIn. scripts/generate_hero.py now runs a five-step ladder: Banana MCP (when the nanobanana-mcp server is loaded) Direct Gemini API via google-genai (when GOOGLE_AI_API_KEY is set) Premium stock APIs (Unsplash, Pexels, Pixabay) - any key suffices Openverse public API (CC-licensed, no key required) Block with setup instructions if none of the above succeed The block is intentional. A missing hero is a missing hero. No silent fallback to a placeholder gray rectangle that ships to production and embarrasses you on the open graph preview a week later. v1.9.1: The 8-Agent Cybersecurity Audit Pass The Blog Delivery Contract introduced new attack surface. generate_hero.py fetches remote images. blog_preflight.py writes provenance files into draft directories. The reviewer reads files the user created. All of that is new pipe between trusted code and the open internet. So v1.9.1 ran 8 specialist agents against the v1.9.0 surface: OWASP Top 10:2021, CWE Top 25:2024, MITRE ATT&CK pattern matching, supply chain analysis, credential handling review, IaC drift check, AI-generated code audit, and threat-model fitness. Composite score against the audit: 90 of 100. Every HIGH and MEDIUM finding closed in v1.9.1. What Got Closed ID Severity Fix VULN-801 / CHAIN-801 HIGH (CRITICAL on cloud) SSRF guard in generate_hero._http_get: scheme allowlist, private/loopback/link-local refuse via ipaddress, 25 MB cap, no-redirect opener. Closes the AWS-IMDS-via-og:image attack path. VULN-802 HIGH Code-enforced iteration counter at <draft>/.iteration-count. MAX_ITERATIONS = 3, exit code 2 on exceed. Was orchestrator-prose-only. VULN-401 HIGH uv.lock with 142 hash-pinned packages. Reproducible install via uv sync --frozen. VULN-803 / CHAIN-802 MEDIUM Nonce-bound reviewer provenance. _init_review_nonce writes secrets.token_hex(16), Gate 4 verifies review.md echoes the nonce. Closes counterfeit reviewer pass. AUTH-001 MEDIUM OAuth client_secret split from token blob. Re-read on construction from oauth_client_path. VULN-804 / VULN-002 MEDIUM / LOW _PreflightNoRedirectHandler refuses non-http(s) schemes inline and refuses 30x redirects on Gate 5 link checks. VULN-404 / 405 / 406 / 411 MEDIUM x3 + INFO Tightened upper bounds on weasyprint, google-genai, sentence-transformers. setuptools floor at 70.0.0 (CVE-2024-6345 RCE). VULN-IAC-001 MEDIUM SHA-256 hashes for install.sh and install.ps1 published in README. Verify integrity before piping curl into bash. VULN-S01 LOW API-key echo in setup_image_mcp.py --project masked as AIza****WXYZ. VULN-IAC-003 LOW install.sh validates skill_name against ^[a-z0-9-]+$ before mkdir / cp. Defense in depth against path-traversal in a tampered repo. The AUTH-002 PKCE finding stayed open as not-in-scope. Installed-app OAuth flow with explicit client_secret.json does not require PKCE; CSPRNG state plus fixed loopback redirect cover the CSRF case. Quality Gates: Final State Tests: 160 to 187 (+27 targeted at the fixes) Prose hygiene: 0 violations Plugin manifest: valid Version coherence: all 14 surfaces at 1.9.1 (4 canonical plus 10 sub-skill SKILL.md) Dependency lock: present (uv.lock, 142 packages, SHA-256 pinned) Projected security composite: ~95 of 100 (up from 87 pre-fix) What This Actually Looks Like in Production The clearest answer is the Codex vs Claude sample blog. That post is the output of one end-to-end run on the v1.9.1 contract. Hero image generated via the ladder. SVG charts inline. JSON-LD validated. FAQ schema present. Score 92 of 100 on the blocking reviewer. Total elapsed time including research, drafting, image generation, rendering, and the full 5-gate pass: about 10 minutes. The folder it produced has the contract artifacts in it: capabilities.json, .iteration-count, .review-nonce, preflight-report.json, review.md. The same gates that produced that post produce every post. Watch the Walkthrough Two ways to consume v1.9.1. Flip through the 12-slide release deck for a 60-second skim, or watch the 12:48 walkthrough below for the full demo, including 6 minutes of live agent rendering. Open the 12-slide release deck → 12-SLIDE DECK · ←→ navigate · F fullscreen · S presenter · A autoplay · ? help Open in new tab ↑ The video shows the contract running live: cover image generation, the SVG chart pipeline, the inline embed flow, the 6-minute silent stretch in the middle is the agent rendering the Codex vs Claude post in real time. Skip to 8:17 if you want to jump straight to the install walkthrough and the finished sample. Upgrade Guidance If you are on v1.7.x or earlier, this is the right upgrade. Read the security FAQ first (links in the v1.7.1 post). Rotate any Google AI API keys that might have been committed to .mcp.json before v1.7.1. Then update. If you are on v1.8.x, update straight through. The two soft-compat shims in v1.9.1 mean nothing breaks if your reviewer config is older. The shims emit a deprecation warning and will be removed in v1.10.0, so update your oauth_client_path config and let your draft regenerate the .review-nonce file once. If you build CI around the skill, pin to v1.9.1 in your install command and use uv sync --frozen against the committed uv.lock for reproducible installs. The 142 hash-pinned packages mean PyPI mirror compromises and typosquats cannot land malicious code at install time. Install or Update curl -sSL https://raw.githubusercontent.com/AgriciDaniel/claude-blog/main/install.sh | bash Verify installer integrity against the SHA-256 hashes published in the README before piping curl into bash. Reproducible alternative via git clone plus uv sync --frozen is documented in docs/INSTALLATION.md. Conclusion v1.8.x taught a lesson. Seven hardening waves in 48 hours, each consuming the previous round's audit output as the next round's input, lifted the security baseline from "good enough" to "audited and locked." v1.9.0 applied that lesson one layer up: from the project's own code to the project's output. v1.9.1 then re-ran the security playbook against the expanded surface. Both shipped together because they are one shift. Audits ladder you up to your asymptote. Infrastructure raises the asymptote. The reviewer is no longer the last advisory voice. It is the gate. // AUTHOR Daniel Agrici AI Automation Specialist based in Chisinau, Moldova. Creator of 21 open-source repositories with 4,900+ GitHub stars total, including Claude Blog and Claude SEO. GovTech Hackathon Moldova first-place winner. Writes about evidence-led AI content workflows, security hardening of AI tooling, and the operator's view of running Claude Code in production. Follow: GitHub // YouTube // Skool // LinkedIn // FAQ FREQUENTLY ASKED QUESTIONS What is the Blog Delivery Contract introduced in v1.9.0? - The Blog Delivery Contract is a 5-gate pipeline that runs before any draft reaches you. Gate 1 enumerates capabilities, Gate 2 checks format completeness (md + html + pdf + real hero), Gate 3 runs Playwright at 375 / 768 / 1280 widths, Gate 4 is the BLOCKING content reviewer (score 90+ or no delivery), Gate 5 verifies every asset and link. The reviewer used to be advisory. Now it blocks. What does BLOCKING mean for Gate 4 in practice? + BLOCKING means the orchestrator does not return the draft if the blog-reviewer agent scores it below 90 of 100 or surfaces any P0 issue. The orchestrator iterates up to 3 times against the gates before escalating to the user with a diagnostic. The user is no longer the first reviewer. The gates are. What did the v1.9.1 security audit find and fix? + An 8-agent cybersecurity audit against OWASP Top 10:2021, CWE Top 25:2024, and MITRE ATT&CK surfaced 11 findings. v1.9.1 closes every HIGH and MEDIUM: SSRF guard in hero generation (VULN-801, closes the AWS-IMDS-via-og:image attack path), code-enforced iteration counter (VULN-802), nonce-bound reviewer provenance (VULN-803), uv.lock with 142 hash-pinned packages (VULN-401), no-redirect HTTP handler (VULN-804), tightened dependency upper bounds, OAuth client_secret split from token blob (AUTH-001), API-key echo masking, and skill_name regex defense-in-depth on the installer. Test suite is now 187 tests, up from 160. Does v1.9.1 break backward compatibility? + No. Two soft-compat shims keep older flows working through v1.9.x. First, the nonce-bound reviewer provenance soft-passes with a deprecation warning when the .review-nonce file is absent. Second, the OAuth path falls back to legacy token-file client_secret if oauth_client_path is unset. Both shims will be removed in v1.10.0, so update your reviewer config and re-authorize OAuth to silence the warnings now. What is the Codex vs Claude sample blog you keep mentioning? + It is a real post produced end-to-end by the v1.9.x gates as proof of skill. Topic: ChatGPT Codex vs Claude Code, who wins in 2026. Generated in about 10 minutes including research, hero image, SVG charts, FAQ, and a 5-gate review pass. Read it here to see what the contract actually produces, not what the docs claim it produces. // NEW FROM CLAUDE BLOG CLAUDE BLOG V2.1.1 RELEASE Continue with the Google guidance, honest scoring, dependency, and public-release safeguards in v2.1.1. READ ARTICLE → // RELATED KEEP READING CODEX VS CLAUDE: WHO WINS IN 2026? The sample blog the v1.9.1 gates produced end-to-end during the walkthrough. Hero, charts, FAQ, schema, 92/100 review pass, ~10 minutes. READ ARTICLE → CLAUDE BLOG V1.9.1 SECURITY HARDENING The predecessor security pass: 1 CRITICAL plus 5 HIGH closed in 24 hours. Sets the precedent for the v1.9.1 audit cycle. READ ARTICLE → AI CONTENT PIPELINE GUIDE End-to-end content workflow on Claude Code. Where the Delivery Contract sits in a production pipeline. READ ARTICLE → RELEASE ON GITHUB v1.9.1 cumulative release notes covering all 9 versions from v1.8.0 through v1.9.1. Install, verify SHA-256, run. VIEW RELEASE → UPDATE TO V1.9.1 TODAY. $ curl -sSL https://raw.githubusercontent.com/AgriciDaniel/claude-blog/main/install.sh | bash COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## Claude Code vs Codex: Who Should Review Whom? | Claude Blog URL: https://claude-blog.md/blog/chatgpt-codex-vs-claude-code-2026 Claude Code vs Codex: Who Should Review Whom? | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/Claude Code vs Codex Who Should Review Whom // UPDATED UNDER CLAUDE-BLOG QUALITY GATES · This comparison was refreshed with current official documentation, evidence-led review rules, link checks, schema validation, and editorial review. See how the gates work → // COMPARISON CLAUDE CODE VS CODEX WHO SHOULD REVIEW WHOM? DANIEL AGRICI // PUBLISHED MAY 20, 2026 // UPDATED AUG 4, 2026 // 10 MIN READ // COMPARISON AI CLAUDE CODE CODEX Key Takeaways • The best first reviewer is usually the model that did not write the change. • Keep review read-only, require file-level evidence, and let the original writer own fixes. • Stop after two review passes. Tests, CI, and an accountable human remain the release authority. Claude Code should review Codex when Codex wrote the change. Codex should review Claude Code when Claude wrote it. That is the practical default. The point is not to crown one permanent winner. It is to separate implementation from critique so the reviewer approaches the diff without defending its own earlier choices. This is a workflow recommendation, not proof that two models always beat one. A second model can catch a missed edge case, but it can also repeat the same assumption or invent a false positive. The value comes from independence, constrained permissions, explicit evidence, and a stopping rule. The product names will keep changing. Model releases, benchmark claims, and plan pricing expire too quickly to serve as durable selection rules. This guide relies on current official workflow documentation and avoids a fixed model ranking. Editorial note: This page was updated and fact-checked on August 4, 2026 against the current official OpenAI and Anthropic documentation linked below. Quick Answer: Pick the Reviewer by Authorship Situation First reviewer Why Claude Code wrote the diff Codex Independent review is directly supported by Codex review tools. Codex wrote the diff Claude Code A restricted Claude reviewer can inspect the change in a separate context. Human wrote the diff Either, then switch if needed Choose the reviewer with the clearest repository context and domain instructions. High-risk change Both, sequentially Use distinct scopes, then escalate disagreements to a human owner. Small, low-risk change One independent reviewer More agent passes can cost more than the risk justifies. Authorship is only the default. If one tool has the repository instructions, test environment, or specialist context the other lacks, that evidence can justify a different order. Write the exception down so the workflow stays auditable. Why Independence Matters More Than a Model Ranking Review independence means the reviewer receives the request, diff, checks, and project rules without inheriting the writer's need to defend its own implementation. Review is a different job from implementation. The writer tries to satisfy the request and make the checks pass. The reviewer tries to falsify that result: find an unhandled input, a contract violation, an accidental behavior change, or a test that proves less than it appears to prove. OpenAI documents a dedicated Codex review flow that reads a selected diff and returns prioritized findings without changing the working tree. It can review changes against a base branch, uncommitted work, a commit, or a custom scope. See the official Codex code review documentation. Anthropic documents Claude Code subagents with their own context, system prompt, tools, and permissions. Its examples include read-only exploration and code-review roles. See the official Claude Code subagent documentation. Those capabilities make a separated reviewer possible in either direction. They do not guarantee a correct verdict. One writer, one independent reviewer, one accountable release gate. The Five Quality Gates 1. Scope Gate: Freeze the Review Target Give the reviewer one explicit target: the current uncommitted diff, one commit, or the branch against a named base. Include the acceptance criteria and repository instructions. Do not ask for a vague review of the entire codebase unless that is truly the task. A useful scope statement is: Review this branch against main for correctness, security, regressions, and missing tests. Do not edit files. The same discipline powers a strong AI content pipeline: each gate has a defined input, output, and failure condition. 2. Baseline Gate: Run Deterministic Checks First Run the closest tests, type checks, lint rules, and build before paying a model to infer what the tools can prove. Record any pre-existing failure. Otherwise the reviewer may attribute old breakage to the new diff, or spend its context describing a compiler error you already had. 3. Evidence Gate: Reject Unsupported Findings Every actionable finding should name the file and location, explain the failure mode, identify the affected input or caller, and recommend the smallest safe correction. A preference such as "I would structure this differently" is not a defect unless it violates an explicit project rule. This is the coding equivalent of the evidence checks in AI content scoring. Severity labels are useful only when they change the decision. High means block release. Medium means fix or consciously accept. Low means optional improvement. 4. Fix Ownership Gate: Return Findings to the Writer The reviewer should not silently repair its own findings. Send valid findings back to the original writer, because that agent holds the implementation context and can update tests with the fix. Then rerun the deterministic checks before the second review pass. 5. Stop Gate: Cap the Loop Stop after at most two review passes when no new high-severity evidence-backed finding appears and all required checks pass. If the models disagree, capture both claims and ask the human owner to decide. Never let two agents debate until one yields. When Claude Code Writes, Let Codex Review Read-only review is a review pass that returns evidence-backed findings without modifying the working tree. This direction has the most direct bridge. OpenAI maintains the Codex plugin for Claude Code. Its /codex:review command performs a normal read-only review, while /codex:adversarial-review supports a more challenging, steerable review for riskier decisions. Ask Claude Code to implement one scoped change and run the required checks. Inspect the diff and confirm there are no unrelated files or secrets. Run the Codex review against that diff with the acceptance criteria attached. Have Claude Code address only evidence-backed findings. Run checks again, then allow one final Codex pass if the fixes were material. The plugin also offers an automatic review gate. Its own documentation warns that a stop hook can create long-running Claude and Codex loops and drain usage. Use that gate only with a hard pass limit and a clear failure handoff. A scheduled automation needs the same protection. When Codex Writes, Let Claude Code Review Fix ownership means the original writer remains responsible for applying accepted findings and rerunning the deterministic checks. Use a dedicated Claude Code reviewer subagent or a plan-only session. Deny edit and write tools. Give it the same diff, acceptance criteria, project instructions, and baseline results. Anthropic's CLI reference documents plan permission mode and tool restrictions, while the subagent documentation shows how to define a specialist reviewer. ROLE: Independent code reviewer. SCOPE: Current branch against main. PERMISSIONS: Read-only. CHECK: Correctness, security, regressions, missing tests. RETURN: Severity, file and line, evidence, smallest safe fix. IGNORE: Style preferences that do not violate repository rules. Do not provide the writer's hidden chain of reasoning or ask Claude to validate Codex's confidence. Give it the artifacts that matter: request, diff, tests, logs, and explicit constraints. Independent context is the feature. When Both Models Should Review Two sequential reviewers can be justified for authentication, billing, permissions, destructive migrations, public API compatibility, or large dependency changes. Split their assignments so they do not simply echo each other. Reviewer scope Questions to answer Correctness reviewer What input, branch, or state transition produces the wrong result? Security reviewer What trust boundary, permission, secret, or destructive action changed? Regression reviewer Which existing consumer or contract can the diff break? Test reviewer Which claim is untested, and what assertion would prove it? Run them sequentially or in separate worktrees. Never allow simultaneous writes to the same checkout. Separate worktrees make attribution and rollback clear, while a single fix owner prevents two partial solutions from colliding. What Cross-Model Review Cannot Replace Tests: a plausible explanation is not executable evidence. Static analysis: a model should not duplicate deterministic rules. Repository policy: both tools must follow the same checked-in instructions. Human accountability: a person still owns risk acceptance and release. Production observation: monitoring and rollback remain necessary after merge. If you want reusable reviewer instructions, start with a small skill rather than a giant universal prompt. The Claude Code skills guide explains how to package focused workflows. For an example of hard quality gates, see the Claude Blog delivery contract. Final Verdict Use the other model as the first reviewer. Claude Code reviews Codex output. Codex reviews Claude Code output. For high-risk changes, use both sequentially with different scopes. Keep review read-only, require evidence, return fixes to one owner, and stop after two passes. That workflow survives the next model release because it does not depend on a temporary benchmark lead. It depends on separation of duties, deterministic checks, and an explicit human decision at the end. // AUTHOR About the Author: Daniel Agrici Daniel Agrici is an AI Automation Specialist based in Chisinau, Moldova. He builds open-source systems for AI content workflows, code review, SEO, and knowledge management. He writes about evidence-led automation and production guardrails. Follow his work on GitHub and YouTube, join the AI Marketing Hub, or contact Daniel through LinkedIn. // FAQ FREQUENTLY ASKED QUESTIONS Should Claude Code review Codex or should Codex review Claude Code? - Use the model that did not write the change as the first reviewer. That separation matters more than a universal ranking. Keep the original writer responsible for fixes, then let tests, CI, and a human owner resolve any remaining disagreement. Can I run Claude Code and Codex on the same repository? + Yes, but run them sequentially for the same change or isolate them in separate worktrees. Do not let both agents write to the same checkout at the same time because overlapping edits make attribution and rollback harder. How do I make cross-model review read-only? + Use Codex review mode or the OpenAI plugin's review command when Codex is the reviewer. For Claude Code, use a reviewer subagent or plan mode with edit and write tools denied. The reviewer should return findings, not modify the working tree. Does cross-model review guarantee fewer bugs? + No. A second model can provide a different review perspective, but it can also miss defects or raise false positives. Tests, static analysis, CI, and accountable human judgment remain the release authority. When should a Claude Code and Codex review loop stop? + Stop after no more than two review passes when no new high-severity evidence-backed finding appears and the required checks pass. Record unresolved disagreements and send them to the human owner instead of allowing an open-ended agent loop. // RELATED KEEP READING V1.9.1: 5 GATES BLOCK SLOPPY DRAFTS The release announcement behind this article's evidence, review, schema, and link-integrity gates. READ ARTICLE → BEST CLAUDE CODE SKILLS 2026 A roundup of the highest-leverage skills for working in Claude Code, including the blog and SEO stack referenced here. READ ARTICLE → ALL CLAUDE BLOG SKILLS The focused skills behind the contract: writers, reviewers, schema, FAQ, images, link integrity, and more. VIEW SKILLS → STAR CLAUDE-BLOG ON GITHUB MIT-licensed and free forever. Read the source, file issues, or star the repo to follow along with future releases. VIEW REPO → RUN THIS YOURSELF IN ABOUT 10 MINUTES. $ curl -sSL https://raw.githubusercontent.com/AgriciDaniel/claude-blog/main/install.sh | bash COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## claude-blog v1.9.1 · Release Deck URL: https://claude-blog.md/decks/v1-9-1/ claude-blog v1.9.1 · Release Deck SYS · CLAUDE-BLOG · V1.9.1 · ACTIVE RELEASE · v1.9.1 · 2026-05-18 claude blog. The Content Operating System. Drafts that ship gated, audited, and production-grade, or do not ship at all. LIVE 5 gates 187 tests 95/100 MIT ★ Lesson 1 · The 5-Gate Method What you’ll learn in 2 minutes M2The 5-gate Blog Delivery Contract M4How to make a reviewer BLOCK, not advise M5Hero image fallback ladder pattern M6Closing OWASP / CWE findings end-to-end G1 discover G2 format G3 visual G4 review G5 links $ /blog write “why your AI blog tool is shipping broken drafts” generating hero · banana mcp drafting Engineering · 7 min read By Daniel Agrici · 2026-05-18 · gated by /blog write security score · v1.8 → v1.9.1 G1 … G2 … G3 … G4 … G5 … Finally. Drafts that don’t ship broken. CLAUDE-BLOG · DELIVERY CONTRACT + AUDIT PASS · v1.9.0 → v1.9.101 / 12 SLIDE 02 · WHY THIS RELEASE 02 · The gap we filled AI blog tools ship sloppy drafts. The user becomes the reviewer. Three structural defects show up over and over in AI-authored content. v1.9 closes all three at the gate layer, before any draft reaches you. SLOP 01 Broken hero images no fallback ladder · 404 og:image Most tools point at a stock URL and pray. When the link 404s the draft ships anyway. Visible the first time anyone shares it. SLOP 02 No PDFs, no real renders markdown-only · .md / .html drift One format out. No print, no offline copy. Diagrams overflow on mobile. The HTML and markdown drift quietly apart. SLOP 03 Reviewer scored, never blocked advisory-only · prose feedback A reviewer step that emits a score nobody reads. Sloppy drafts pass anyway because there is no gate, just a comment. …this can’t ship. CLAUDE-BLOG · WHY THIS RELEASE MATTERS02 / 12 SLIDE 03 · WHAT SHIPS 03 · What shipped in 1.9 Two releases, one outcome, five load-bearing surfaces. v1.9.0 introduced the delivery contract. v1.9.1 hardened it against an eight-agent cybersecurity audit. Every number below is verifiable in the repo. 5 delivery gates NEW IN v1.9 187 pytest tests +27 NEW 29 sub-skills 95 security / 100 87 → 95 142 pinned packages CLAUDE-BLOG · 5 · 187 · 29 · 95 · 14203 / 12 SLIDE 04 · DELIVERY CONTRACT 04 · The Blog Delivery Contract Five gates run before any draft reaches you. No exceptions. Each gate is code, not prose. Gate 4 is the blocking one. The other four make sure Gate 4 has something real to score. GATE 1 Capability discovery MCPs · agents · scripts Write capabilities.json. Know what is loadable before drafting. GATE 2 Format completeness .md · .html · .pdf · hero Every draft ships four artifacts. Deterministic renderer. GATE 3 Visual verification 3 widths · overflow · JSON-LD Playwright renders. Asserts no SVG overflow. Captures console errors. GATE 4 Content review BLOCKING · 90+ · zero P0 Reviewer agent scores rendered HTML. Below threshold, no delivery. BLOCKING GATE 5 Asset + link integrity img src · 200 OK · wordCount Every link resolves, every image exists, JSON-LD matches body. Gate 4 is the one that bites. CLAUDE-BLOG · 5-GATE DELIVERY CONTRACT · v1.9.004 / 12 SLIDE 05 · VISUAL VERIFICATION 05 · Gate 3: visual verification Mobile breaks fail at the gate. Not in your face. Playwright spins up a headless browser and renders the HTML at three widths. It asserts every SVG child fits inside its viewBox, swaps to dark mode, and watches the console for errors. claude-blog.dev/article 375 px Why your AI blog tool is broken ✓ SVG within viewBox ✓ dark-mode swap ✓ no overflow-x claude-blog.dev/article 768 px Why your AI blog tool is shipping broken drafts ✓ hero crop center ✓ pull-quote breaks ✓ TOC sticky claude-blog.dev/article 1280 px Why your AI blog tool is shipping broken drafts (and how to fix it) ✓ sidebar layout ✓ code-block scroll ✓ social card OK CLAUDE-BLOG · GATE 3 · PATCHRIGHT · 3 WIDTHS05 / 12 SLIDE 06 · BLOCKING REVIEWER 06 · Gate 4: the blocking one Score 90+. Zero P0. Or no delivery. The blog-reviewer agent scores the rendered HTML against six dimensions. Threshold is 90 of 100 with zero P0 issues. Below that, the orchestrator iterates, up to three times. REVIEW.MD · 6 DIMENSIONS /blog write · Why your AI blog tool is broken Hook + answer-first opener17 / 18 Heading hierarchy + skimmability14 / 15 Citations + tier-1 sourcing22 / 22 Original perspective + E-E-A-T19 / 20 AI-citation readiness (FAQ + schema)13 / 15 No P0 editorial defects10 / 10 TOTAL · THRESHOLD 90 95 BLOCKING: false P0 ISSUES: 0 Nonce: 4f2a8c11d6e93b7a · verified Iteration: 1 of 3 · agent: blog-reviewer 95 / 100. Let it ship. CLAUDE-BLOG · GATE 4 · BLOCKING TRUE | FALSE06 / 12 SLIDE 07 · HERO LADDER 07 · Every draft ships a real hero Five-step fallback ladder. Block if none work. Hero generation tries banana first, falls back through Gemini, then premium stock, then openverse. If nothing returns a real image, the gate blocks delivery with setup steps. No silent 404s. 01 Banana MCP orchestrator level · when nanobanana-mcp is loaded MCP 02 Gemini API direct google-genai · when GOOGLE_AI_API_KEY is set API 03 Premium stock APIs Unsplash · Pexels · Pixabay · any key suffices STOCK 04 Openverse public API CC-licensed · no key required · always available CC 05 Block with setup steps if none succeed · honest diagnostic to the user BLOCK Never a silent 404. CLAUDE-BLOG · GENERATE_HERO.PY · 5-STEP FALLBACK07 / 12 SLIDE 08 · CYBERSECURITY AUDIT 08 · Then we ran the audit Eight cybersecurity agents. Eleven findings. All closed. OWASP Top 10:2021, CWE Top 25:2024, MITRE ATT&CK mapping. Each agent scored independently. Composite score went from 87 to 95 of 100 after the v1.9.1 hardening pass. A1 Vulnerability OWASP Top 10 + CWE Top 25 + CVE database A2 Secrets hardcoded keys · committed creds · token leaks A3 Dependencies supply chain · CVE bounds · typosquats A4 IaC install scripts · permissions · path traversal A5 Threat intel malware · backdoor · C2 · MITRE ATT&CK A6 Authorization OAuth · token storage · session boundaries A7 AI-gen code agent provenance · prompt injection vectors A8 Compliance OWASP coverage · CWE mapping · score band CLAUDE-BLOG · 8-AGENT AUDIT · v1.9.1 HARDENING08 / 12 SLIDE 09 · WHAT CLOSED 09 · What got closed Every HIGH and MEDIUM finding. Closed. Eleven audit findings landed. v1.9.1 closes all HIGH and MEDIUM severity items, plus the LOW and INFO ones worth fixing. Each fix is wired to a regression test so the same class of bug cannot recur silently. VULN-801HIGH SSRF + IMDS attack path scheme allowlist · private IP refuse · 25MB cap VULN-802HIGH Iteration counter unenforced code-enforced .iteration-count · cap = 3 VULN-401HIGH Supply chain non-determinism uv.lock · 142 packages · SHA-256 pinned VULN-803MED Counterfeit reviewer pass nonce-bound review.md provenance · 32-hex AUTH-001MED OAuth client_secret in token blob split secret out · re-read from config VULN-804MED Gate 5 redirect-follow no-redirect opener · scheme check inline VULN-IAC-001MED Installer hashes unpublished SHA-256 published in README · verify step VULN-S01LOW API key echo in setup _mask_api_key helper · AIza****WXYZ style v1.9.1 · shipped "Why your AI blog tool is shipping broken drafts" Tests: 160 → 187 (+27) Security score: 87 → 95 / 100 Composite: ~94 / 100 Band: Exceptional Every HIGH + MED. Closed. CLAUDE-BLOG · v1.9.1 · 11 FINDINGS CLOSED09 / 12 SLIDE 10 · ECOSYSTEM 10 · The full sub-skill map Twenty-nine sub-skills. One orchestrator. Every job. CLAUDE-BLOG · ASSETS/DIAGRAMS/03-SUB-SKILL-MAP-A.SVG10 / 12 SLIDE 11 · ROADMAP 11 · What v1.10 and beyond bring v1.9 is shipped. Nonce-mandatory, AOL, citations are next. CLAUDE-BLOG · v1.9 NOW · v1.10+ NEXT11 / 12 SLIDE 12 · INSTALL v1.9.1 · ready to install One line. Five gates. Ship it. $curl -sSL https://raw.githubusercontent.com/AgriciDaniel/claude-blog/main/install.sh | bash ★ NEXT LESSON Build this for your stack subscribe · new lessons drop weekly INSTALL + PRACTICE Run the 5 gates locally github.com/AgriciDaniel/claude-blog COURSE COMMUNITY Join AI Marketing Hub lessons · weekly threads · 2,700+ members Catch the next one. Same gates. CLAUDE-BLOG · THANKS FOR WATCHING12 / 12 01 / 12 ? for help · O overview · S presenter · A autoplay Keyboard map → / Space / N Next slide ← / P Previous slide 1 – 9 Jump to slide 1–9 0 Jump to last slide Home / End First / Last slide R Replay reveals O Overview grid S Presenter view A Autoplay (12s/slide) F Fullscreen ? Show / hide this help Esc Close overlay Append ?print=1 to the URL for a printable view Slide overview Presenter view SLIDE 01 Cover Open. Banner slams. NEXT Why this release 00:00 since last advance --- ## Claude Blog v1.7.0: 5 Community Skills Pro Hub Release URL: https://claude-blog.md/blog/claude-blog-v170-community-release Claude Blog v1.7.0: 5 Community Skills Pro Hub Release Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/Claude Blog v1.7.0 Community Release // RELEASE UPDATE CLAUDE BLOG V1.9.1 5 COMMUNITY SKILLS DANIEL AGRICI // MAY 11, 2026 // 7 MIN READ // RELEASE COMMUNITY v1.7.0 is the first community-driven release of Claude Blog. Five new sub-skills landed via the AI Marketing Hub Pro Hub Challenge, alongside the FLOW framework integration that turns 30 evidence-led prompts into a built-in workflow. The architecture grew from 22 to 28 sub-skills overnight, the agent roster grew from 4 to 5, and the command surface jumped from 17 to 27, all without compromising the security posture or the quality bar that v1.6.7 set. What is the Pro Hub Challenge? The Pro Hub Challenge is a community program inside the AI Marketing Hub Pro. Pro members were invited to build and submit new skills and extensions for Claude Blog and Claude SEO. Each submission ran through an independent audit covering seven dimensions: security, functionality, code quality, documentation, dependency hygiene, SKILL.md discoverability, and innovation. Scores are out of 100, and a Proficient grade (75 or above) is the minimum threshold for integration consideration. Six submissions reached the audit. Five scored Proficient or above. After security review and a clean-room re-implementation in the Claude Blog voice and security posture, two submissions integrated into Claude Blog v1.7.0. Four submissions in total landed in Claude SEO v1.9.0 (an overlap by design, since some contributors wrote dual-target skills that worked on both the blog and the SEO sides). The remaining submissions were either rejected for security reasons or deferred until the analyzer can be separated from page-builder dependencies. This is the cadence that the program is meant to produce. Pro members build what they need, the audit framework keeps the bar high, and the integrated contributions ship to everyone with the contributor's name on the source. The five community skills Two contributors, five skills. Here is the full roster. blog-cluster (Lutfiya Miller, 95/100 Exemplary) Semantic topic clustering via SERP overlap. Hub-and-spoke architecture, sequential blog-write orchestration with shared cluster context, automatic internal-link injection across hub and spoke posts, and an XSS-hardened cluster-map.html visualization with no inline JavaScript. Lutfiya's design separates Plan from Execute, which means the cluster strategy can be reviewed and adjusted before any post is written. The same contributor's submission also produced seo-cluster for Claude SEO v1.9.0, which handles the SERP-based architecture planning side. This was the highest-scoring submission of the entire challenge. Read more at /skills/blog-cluster. blog-multilingual (Chris Mueller, 85/100 Proficient) One-command international publishing orchestrator. Calls blog-write, blog-translate, blog-localize, and blog-locale-audit in sequence, then emits the matching hreflang tags. Optional seo-hreflang integration ties it back to the Claude SEO side when both packs are installed. The skill is the most blog-native submission of the challenge: four user-facing skills explicitly designed for the Claude Blog publishing pipeline. Read more at /skills/blog-multilingual. blog-translate (Chris Mueller, same submission) Native-quality SEO translation with format preservation. Markdown, frontmatter, schema, and inline SVG survive the round trip intact, so a translated post does not lose its JSON-LD or its visual structure. Locale-correct number, date, currency, and quote formatting are applied automatically. Machine-translation artifact detection runs as a post-pass and flags the kind of awkward phrasing that crosses straight from training data without locale adaptation. Read more at /skills/blog-translate. blog-localize (Chris Mueller, same submission) Cultural deep-adaptation beyond translation. Built-in profiles ship for DACH (Germany, Austria, Switzerland), Francophone, Hispanic, and Japanese markets, each tuned for formality conventions, currency, units, and idiomatic phrasing. Custom profiles are YAML-defined, so adding a new locale (Portuguese-Brazil, Arabic, Korean) is a config change rather than a code change. The shared cultural-adaptation.md reference is treated as a single source of truth and referenced (not duplicated) across both blog-localize and the underlying blog-translator agent. Read more at /skills/blog-localize. blog-locale-audit (Chris Mueller, same submission) Multilingual quality control. Completeness matrix across all language variants flags missing translations, hreflang correctness validation catches the most common i18n bug (asymmetric or self-referential hreflang), meta-tag parity ensures titles and descriptions are present in every locale, and freshness tracking flags translations that have fallen behind the source post after a content update. Read more at /skills/blog-locale-audit. Plus FLOW framework integration Alongside the community skills, v1.7.0 also ships blog-flow. The skill syncs 30 blog-applicable evidence-led prompts (Find: 5, Leverage: 1, Optimize: 21, Win: 3) from github.com/AgriciDaniel/flow under CC BY 4.0. Context-matching orchestration selects two or three prompts per Optimize-stage analysis based on niche and prior skill output, so the framework adapts to the post rather than the other way around. The local-stage prompts (Google Business Profile, citations, local audits) are intentionally excluded; that work belongs in Claude SEO's seo-flow. Read more at /skills/blog-flow. The blog-translator agent In addition to the five user-facing skills, Chris Mueller's submission added the blog-translator agent. It specializes in translation and localization with format preservation, and is the underlying engine the multilingual skills route to for high-volume conversions. The agent ships with no Bash tool grant per the v1.7.1 security baseline. This is a deliberate choice: agents with shell access expand the prompt-injection blast radius on any post that quotes external content. Translation work does not need a shell, so it does not get one. Total agent count rose from 4 to 5: Researcher, Writer, SEO, Translator, Reviewer. Security review process Every Pro Hub submission went through a security audit before integration. Four mechanical guardrails now run as pytest gates on every commit: No agent grants the Bash tool in its frontmatter (prompt-injection blast radius). No SKILL.md includes the invalid allowed-tools field. Skill names are unique across the entire repository (no duplicate routing). The FLOW sync script preserves all six of its security invariants (host allowlist, size cap, dry-run flag, ref pinning, lockfile, license-header injection, path-traversal guard). Submissions that did not clear the bar were rejected or reworked before integration. One submission (the SEO drift monitor) was rejected outright for a hardcoded Google API key. Two of the four submissions that did clear the bar landed in Claude Blog v1.7.0 (Lutfiya's cluster, Chris's multilingual suite), scoring Exemplary and Proficient respectively. Chris's submission also originally shipped a curl | bash installer; both that and the original credential-handling pattern were removed in the port. What this means for the skill catalog The catalog grew, but not at the expense of the bar. Here is the delta. Surface v1.6.8 v1.7.0 Sub-skills 22 28 Agents 4 5 Commands 17 27 Community release is now the standard pattern for high-quality contributions. The Pro Hub Challenge is becoming a regular cadence, not a one-off event, and the next round is already being scoped. How to use the new skills Install Claude Blog (see claude-blog.md) or upgrade to v1.7.0 via /plugin install claude-blog@agricidaniel-blog-tools. Try clustering: /blog cluster plan best-running-shoes. The skill returns a hub-and-spoke plan you can hand off to /blog cluster execute. Try multilingual: /blog multilingual "AI content scoring" --languages es,fr,de,ja. Writes the source post, translates it, localizes it, and emits hreflang in one pass. Try FLOW: /blog flow optimize <url> for evidence-led optimization on an existing post. The skill selects relevant Optimize-stage prompts based on the niche and the current post state. What's next v1.7.1 (already shipped) hardened the v1.7.0 attack surface and locked the security guardrails as mechanical pytest gates rather than convention. v2 will likely open more community surface; the Pro Hub Challenge is becoming a regular cadence, and the audit framework that vetted the v1.7.0 submissions is now reusable for any future contribution. Conclusion v1.7.0 is the first time Claude Blog grew not just from my own work but from the work of contributors who chose to give it away. Lutfiya's cluster engine and Chris's multilingual suite carry their authors' names in CONTRIBUTORS.md and in every skill's source. That is the open-source kernel made literal: a thing one person built that anyone can extend, with the extensions credited and the bar kept. ABOUT THE AUTHOR Daniel Agrici is an AI Automation Specialist based in Chisinau, Moldova. He is the creator of 21 open-source repositories including Claude Blog and Claude SEO, the 1st place winner of the GovTech Hackathon Moldova, and the author of the FLOW framework for evidence-led content workflow. Find him on GitHub. // FAQ FREQUENTLY ASKED QUESTIONS What is the Pro Hub Challenge? - The Pro Hub Challenge is a program inside the AI Marketing Hub Pro community where Pro members build and submit new skills or extensions for Claude Blog and Claude SEO. Submissions are independently audited across security, functionality, code quality, documentation, dependencies, discoverability, and innovation. Submissions that score Proficient or above are eligible for integration after a security review and a clean-room port into the project voice. Who contributed to v1.7.0? + Two contributors landed in Claude Blog v1.7.0. Lutfiya Miller authored the semantic-cluster-engine submission (score 95 of 100, Exemplary) which became blog-cluster. Chris Mueller authored the claude-blog-multilingual submission (score 85 of 100, Proficient) which became four skills: blog-multilingual, blog-translate, blog-localize, blog-locale-audit, plus the blog-translator agent. What is the difference between blog-cluster and seo-cluster? + Both skills come from the same Pro Hub Challenge submission by Lutfiya Miller. blog-cluster ships in Claude Blog and orchestrates blog-write for the per-post execution layer with shared cluster context and automatic internal-link injection. seo-cluster ships in Claude SEO v1.9.0 and is used for SERP-based cluster architecture planning. They are designed to be installed side by side. Why is the agent count now 5 instead of 4? + v1.7.0 adds the blog-translator agent from Chris Mueller's submission. It specializes in translation and localization with format preservation across markdown, frontmatter, schema, and SVG. The agent ships with no Bash tool grant per the v1.7.1 security baseline. The roster is now Researcher, Writer, SEO, Translator, and Reviewer. What is FLOW and where do the prompts come from? + FLOW is an evidence-led content workflow framework (Find, Leverage, Optimize, Win) authored by Daniel Agrici. The blog-flow skill syncs 30 blog-applicable prompts from github.com/AgriciDaniel/flow under CC BY 4.0. The sync script runs stdlib only, host-allowlisted to api.github.com, supports dry-run and ref pinning, and writes a SHA-256 lockfile to detect drift. // RELATED KEEP READING BLOG-FLOW: EVIDENCE-LED WORKFLOW 30 blog-applicable FLOW prompts (Find, Leverage, Optimize, Win) synced from github.com/AgriciDaniel/flow under CC BY 4.0. VIEW SKILL → BLOG-CLUSTER: SEMANTIC TOPIC CLUSTERS Hub-and-spoke topic-cluster planning + execution. SERP-based grouping, sequential write orchestration, XSS-hardened cluster map. VIEW SKILL → BLOG-MULTILINGUAL: ONE-COMMAND I18N Write, translate, localize, audit, hreflang in one pass. Built-in profiles for DACH, Francophone, Hispanic, Japanese. VIEW SKILL → V1.6.6-V1.6.8: 29 ACCURACY FIXES Two independent audits, 88 items reviewed, 29 corrections applied. GA4 quotas, AI crawler specs, schema deprecations. READ ARTICLE → INSTALL V1.7.0 TODAY. FREE AND OPEN SOURCE. $ claude plugin install github:AgriciDaniel/claude-blog COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## Claude Blog v1.7.1: Security Hardening Arc | Claude Blog URL: https://claude-blog.md/blog/claude-blog-v171-security-hardening Claude Blog v1.7.1: Security Hardening Arc | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/Claude Blog v1.7.1 Security Hardening // SECURITY RELEASE CLAUDE BLOG V1.9.1 SECURITY HARDENING ARC DANIEL AGRICI // MAY 11, 2026 // 8 MIN READ // SECURITY CHANGELOG v1.7.1 shipped on the same day as v1.7.0. That is not a coincidence. The v1.7.0 community release triggered a security audit pass that surfaced 1 CRITICAL and 5 HIGH vulnerabilities, and v1.7.1 closed every one of them within 24 hours. This is the story of the security debt that was hiding under v1.7.0's release celebrations, and how it got paid down before anyone noticed it was overdue. The Trigger: Expanded Surface Area, Expanded Threat Model v1.7.0 was the largest community release in the history of the project. New translation API integrations, NotebookLM workflows, Drive credential flows, and contributor-submitted skills doubled the third-party-code surface in a single release. Independently of those contributions, the v1.6.x credential-handling code had quietly accumulated subtle issues. They were the kind of issues that any individual code review would miss, but a focused pre-release audit catches in 20 minutes. The audit ran against the v1.7.0 release candidate. Six findings came out of it: one CRITICAL, five HIGH. Every fix shipped in v1.7.1 the next day. Here is what was wrong and how it got fixed. VULN-001 (CRITICAL): API Key Exposure via .mcp.json Tracking The bug: setup_image_mcp.py wrote the user's Google AI API key to .mcp.json in the user's project directory. .mcp.json is not in the default .gitignore on most setups. If the user ran git add . and committed the file, the key landed in public repo history the moment they pushed. The fix: setup_image_mcp.py now defaults to a --global flag that writes the key to ~/.claude/settings.json with file mode 0600 (owner read and write only, no group, no other). Project-local config is still available behind an explicit --local flag, which prints a warning and a suggested .gitignore entry on every invocation. The default is safe. Who is affected: any user who ran image setup on v1.6.5 through v1.7.0 and committed .mcp.json. The advisory mitigation is rotate the Google AI key, scrub git history (the file may live in old commits even if it is gitignored now), and re-run setup on v1.7.1. VULN-002 (HIGH): OAuth CSRF Without State Token NotebookLM and Google API setup used standard OAuth web flows but did not validate the state parameter on callback. RFC 6749 section 10.12 is explicit: state is the canonical CSRF defense for OAuth. Without it, a malicious page could trigger a callback to localhost:port with attacker-supplied code, and the token-exchange code would proceed without checking that the callback corresponded to a request the user actually initiated. The fix: generate a cryptographically random state token before constructing the auth URL, store it locally, and validate strictly on callback. Mismatches reject the token exchange with a clear error. The local listener is also now bound to 127.0.0.1 instead of 0.0.0.0, which prevents callback hijacking from other machines on the same network. VULN-003 (HIGH): Hash-Pinned Dependencies Across 4 Lock Files Supply-chain risk. pip install trusts whatever PyPI serves at install time. A compromised mirror, a typosquat that ships under a real package name briefly, or a hijacked maintainer account can ship malicious code that runs at install time. The fix: hash-pin every dependency in requirements.txt and 3 nested skill requirements files. CI uses pip install --require-hashes. Hash mismatches now fail loud rather than silently installing whatever PyPI happens to serve. This closes the v1.6.x supply-chain attack surface entirely. VULN-004 (HIGH): PowerShell 5.1 Compatibility Regression Not a vulnerability in the classic sense, but a hardening reliability fix. install.ps1 used the PowerShell 7+ three-argument form of Join-Path, which silently fails on the Windows 10 and Windows 11 default of PowerShell 5.1. Users got truncated installs without errors. The skill appeared to install, then silently broke at first run. The fix: rewrote the installer to use the two-argument Join-Path form plus explicit path validation after every join. Works correctly on PowerShell 5.1 and up across every install path. Verified on a clean Windows 11 image with PowerShell 5.1 only. VULN-005 (HIGH): Frontmatter Completeness Enforcement Discovered during the audit: 15 of 27 user-invokable SKILL.md files were missing one or more of description, argument-hint, or license in frontmatter. Not exploitable directly. But a missing license field creates ambiguous redistribution rights, which is a real legal-exposure issue when contributors fork and bundle skills. The fix: a pytest invariant now requires all three fields on every user-invokable skill. CI blocks merges that violate. Existing skills were brought into compliance in the same PR that introduced the invariant. VULN-006 (HIGH): Credentials File Permissions Token cache files at ~/.cache/claude-blog/*.json were created with the default umask. On most Linux and macOS systems, the default umask produces mode 0644, which is world-readable. OAuth tokens are sensitive credentials. World-readable is the wrong choice. The fix: explicit chmod 0600 after every credential write. Writes are now atomic via tempfile plus shutil.move, which closes the partial-write window where a token file could exist on disk with the wrong permissions for a few milliseconds before the chmod ran. The 24-Hour Turnaround Mechanical guardrails enforced via pytest invariants meant that any security-affecting change had to clear an automated bar before merge. The four invariants that made the turnaround possible: No agent grants Bash. Preserves the prompt-injection containment line. If an agent could shell out, prompt injection escalates to remote code execution. No invalid allowed-tools field. Catches typos that silently grant Bash. A field name typo would silently fall back to default, which is overly permissive. Unique skill names. Prevents shadowing attacks where a malicious contribution registers under the same name as a trusted skill. FLOW sync integrity. SHA-256 lockfile validates before write. Any unauthorized modification of the lockfile fails the invariant. All v1.7.1 fixes passed the invariants on the first try, because the invariants existed before the audit. The lesson: write the guardrails before you need them. A security audit becomes a 24-hour fix sprint instead of a multi-week rewrite when the mechanical bar is already in place. What This Means For Users Update to v1.7.1. Specifically: Rotate any Google API key that may have been committed to git via .mcp.json. Re-run image setup if you used v1.6.5 through v1.7.0. Re-authorize Drive sessions and NotebookLM credentials. The previous tokens used permissive file modes and unprotected OAuth flows. If you use Claude Blog in CI, pin to v1.7.1 or later. Hash-pinned requirements are only effective if you pin the skill version too. Conclusion Security hardening sprints are unsexy. Nobody tweets about CRITICAL-to-zero in 24 hours the way they tweet about a new feature. But the work compounds. v1.7.1 closed 6 findings, and the next release inherits the higher baseline. Mechanical guardrails do the work that humans forget. Community submissions raise the bar on everyone, and the security review is the price of entry, not a punishment. Ship the guardrails first. The audit will find things. Make sure the things it finds can be fixed in a day. // AUTHOR Daniel Agrici AI Automation Specialist based in Chisinau, Moldova. Creator of 21 open-source repositories with 4,900+ GitHub stars total, including Claude Blog and Claude SEO. GovTech Hackathon Moldova first-place winner. Writes about evidence-led AI content workflows, security hardening of AI tooling, and the operator's view of running Claude Code in production. Follow: GitHub // YouTube // Skool // LinkedIn // FAQ FREQUENTLY ASKED QUESTIONS What was VULN-001 and who was affected? - VULN-001 was a CRITICAL API key exposure. setup_image_mcp.py wrote Google AI API keys to .mcp.json inside the user's project directory. Any user running v1.6.5 through v1.7.0 who committed .mcp.json to git leaked their key. Mitigation: rotate the key, scrub git history, and re-run setup on v1.7.1 which defaults to the global config path with mode 0600. How does OAuth state token validation work? + Before redirecting the user to the auth URL, v1.7.1 generates a cryptographically random state token and stores it locally. The state is included as a query parameter in the auth URL. When the provider redirects back to the localhost listener, the callback compares the returned state against the stored value. Mismatches reject the exchange. This is the RFC 6749 section 10.12 defense against CSRF. What are hash-pinned dependencies and why do they matter? + Hash-pinned dependencies record the SHA-256 of each package alongside its version in requirements files. pip install --require-hashes refuses to install anything that does not match. This blocks supply-chain attacks via compromised PyPI mirrors, typosquats, or hijacked maintainer accounts. v1.7.1 hash-pins every dependency across 4 lock files. Do I need to update if I am on v1.6.x? + Yes. Several of the fixes apply retroactively to v1.6.5 through v1.7.0. If you ran image setup on any of those versions, rotate your Google AI API key and clean git history if .mcp.json was ever committed. Update via claude plugin install github:AgriciDaniel/claude-blog and re-authorize Drive and NotebookLM credentials. How were the four pytest invariants chosen? + Each invariant maps to a real prompt-injection or shadowing attack. No agent grants Bash preserves containment. No invalid allowed-tools blocks typos that silently widen permissions. Unique skill names prevents shadowing attacks. FLOW sync integrity uses a SHA-256 lockfile to validate every write. The invariants existed before the audit, which is why v1.7.1 fixes passed on the first try. // RELATED KEEP READING CLAUDE BLOG V1.9.1 COMMUNITY RELEASE The community contributions release that triggered the security audit. Translation APIs, NotebookLM, Drive workflows. READ ARTICLE → BLOG FLOW: EVIDENCE-LED CONTENT FLOW framework integration with SHA-256 lockfile sync validation. The fourth pytest invariant in production. READ ARTICLE → BLOG GOOGLE: API INTEGRATION OAuth-secured Google API integration with state token validation, hash-pinned deps, and credentials at 0600. READ ARTICLE → SECURITY.MD ON GITHUB The full security model, threat model, and vulnerability disclosure process for Claude Blog. READ DOCUMENT → UPDATE TO V1.7.1 TODAY. $ claude plugin install github:AgriciDaniel/claude-blog COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## Multilingual Blog Publishing Guide | Claude Blog URL: https://claude-blog.md/blog/multilingual-blog-publishing-guide Multilingual Blog Publishing Guide | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/Multilingual Blog Publishing Guide // TUTORIAL MULTILINGUAL BLOG PUBLISHING ONE COMMAND, EVERY LOCALE DANIEL AGRICI // MAY 11, 2026 // 8 MIN READ // MULTILINGUAL TUTORIAL Most blogs that try to go multilingual fail at the second post. The friction stack, translate, localize, hreflang, QA, deploy, is too tall to maintain across four languages weekly. blog-multilingual collapses that stack into one command. This guide walks you through the full pipeline. Why multilingual blogging usually fails The failure pattern is consistent across teams of every size. It is rarely the translation itself. It is the support work around the translation that wears the cadence down: Translations drift from the source. The English post is updated in March, the German version still reflects the January wording in June. Hreflang tags break when one language gets republished out of order. The return-link reciprocity fails silently and Google quietly stops surfacing the localized variant. Cultural references stay US-centric in translation. A Thanksgiving anecdote in a German audience post lands flat or worse. Meta description length goes over 160 characters in one language but not the others. German runs long. Japanese runs short. Nobody checks until a manual audit, which never gets scheduled. Schema markup loses translated fields after a refactor. Suddenly only the English BlogPosting has a valid headline. Internal links inside translated posts still point at the English versions of the linked pages. These problems compound. Most teams give up by post 3. The one-command flow Here is the full multilingual pipeline as a single invocation: /blog multilingual "AI content scoring explained" --languages es,fr,de,ja What that single command does, in order: Writes the English source post via blog-write, with template auto-selection, sourced statistics, FAQ schema, and the rest of the standard pipeline. Calls blog-translate in parallel for each target language, preserving Markdown, frontmatter, schema, and code formatting. Calls blog-localize in parallel for cultural adaptation per region. Calls blog-locale-audit for QA across all variants. Emits hreflang tags, a sitemap fragment, and a JSON language map. Five posts come out of one invocation: en plus es, fr, de, ja, plus the hreflang stubs ready to paste into your CMS. What gets preserved during translation blog-translate is a format-preserving translator, not a string-level one. It keeps: Markdown structure: heading hierarchy, list indentation, table cell alignment. YAML frontmatter keys (only values are translated, keys stay English). JSON-LD schema blocks, with inLanguage set per target. Inline SVG (text in <text> elements is translated, geometry is untouched). Code blocks (left as-is, never translated). Internal anchor structure, with anchors mapped to the target language slug. Locale-correct formatting is then applied per target. Number separators flip from 1,234.56 to 1.234,56 for German. Dates flip from MM/DD/YYYY to DD.MM.YYYY. Currency converts from USD to EUR at the current exchange rate. Quote conventions flip from "" to „" or « » depending on the language. What localize does that translate does not Translation handles language. Localization handles culture. blog-localize runs after translation and applies a cultural adaptation layer: US idioms swap to equivalent local idioms. "Hit a home run" becomes Volltreffer in German. Examples get replaced. A Super Bowl reference becomes a Bundesliga reference for DACH audiences. Currency is converted, not just symbol-swapped. Imperial units convert to metric. Cultural taboos are flagged and avoided per region. Four cultural profiles ship by default: DACH (Germany, Austria, Switzerland), Francophone (France, Belgium, Quebec), Hispanic (Spain plus Latin America), and Japanese. Custom profiles are defined in YAML so you can add Brazilian Portuguese, Polish, or anything else your audience demands. Hreflang generation Hreflang is where most do-it-yourself multilingual setups break. blog-multilingual emits bidirectional tags where every variant references every other variant plus an x-default. Output is delivered as ready-to-paste HTML and as a sitemap.xml fragment, depending on what your CMS prefers: <link rel="alternate" hreflang="en" href="https://blog.example.com/post" /> <link rel="alternate" hreflang="es" href="https://blog.example.com/es/post" /> <link rel="alternate" hreflang="fr" href="https://blog.example.com/fr/post" /> <link rel="alternate" hreflang="x-default" href="https://blog.example.com/post" /> Every page references all alternates including itself. Every relationship is reciprocal. x-default points to the source-language version. Protocol and trailing-slash conventions stay consistent across the set. Locale-audit as ongoing QA blog-multilingual runs blog-locale-audit once at publish time. Content drifts after that. The recommended cadence is to run the audit weekly across your content directory: /blog locale-audit ./content/ The audit reports three classes of issue: completeness gaps (English post updated, German translation stale), hreflang breakage (a return tag missing after a slug change), and meta parity (a target-language title that runs over 60 characters when the English fits under). Cost model Translation and localization use the Gemini API. The free tier supports tens of thousands of characters per day, enough for most weekly publishing schedules. Premium tier is roughly $0.04 per 1000 tokens for translation work. A typical 1500-word post in four languages costs about $0.30 to translate and localize end to end. Compare that to human translators at $0.10 to $0.30 per word: a 1500-word post across four languages is $150 to $450 in human-translator costs alone, before localization, hreflang setup, or QA. End-to-end example Walking through producing a four-language version of an existing post: Run /blog multilingual "topic" to start from a topic, or /blog translate ./content/post.md --to es,fr,de,ja when the source post is already written. Inspect the outputs in ./content/<locale>/post.md for each target. Run /blog locale-audit ./content/ to verify completeness and hreflang. Deploy. The deploy step is CMS-specific: WordPress via the Polylang plugin, Astro via the i18n routing config, Next.js via App Router locale segments. WordPress users will paste the hreflang HTML into a Polylang language switcher template. Astro users add the language codes to astro.config and let the i18n routing handle URL prefixes. Next.js users place files under app/[locale]/blog/ and reference the JSON language map from the layout. Conclusion The goal is a sustainable multilingual blog cadence, not a one-time stunt. blog-multilingual makes the cost of an extra language a flag, not a hire. Run it on your next post and see how the friction stack collapses. About the author Daniel Agrici is an AI automation specialist based in Chisinau, Moldova. Creator of Claude Blog and 20+ open-source repositories with 4,900+ combined GitHub stars. GovTech Hackathon Moldova 1st place winner. Find him on LinkedIn, YouTube, and GitHub. // FAQ FREQUENTLY ASKED QUESTIONS How is this different from running Google Translate? - Google Translate is a string-level service. blog-multilingual preserves Markdown structure, YAML frontmatter, JSON-LD schema blocks, inline SVG, and code fences. It also runs blog-localize after translation to swap cultural references, convert units and currency, and adapt idioms per region. Finally it emits bidirectional hreflang tags and a CMS-ready language map. Google Translate produces a translated string. blog-multilingual produces a publish-ready multilingual blog post. What formats are preserved in translation? + Markdown structure (headings, lists, tables), YAML frontmatter, JSON-LD schema blocks, inline SVG, code blocks, and internal anchor links. Locale-correct formatting is applied per target: number separators (1,234.56 vs 1.234,56), date formats (MM/DD/YYYY vs DD.MM.YYYY), currency conversion, and quote conventions per language. Which cultural profiles ship by default? + Four built-in profiles: DACH (Germany, Austria, Switzerland), Francophone (France, Belgium, Quebec), Hispanic (Spain, Latin America), and Japanese. Custom profiles can be defined in YAML. Each profile covers idiom substitution, example replacement, unit conversion, currency conversion, and cultural taboos to avoid. How are hreflang tags generated? + blog-multilingual emits bidirectional hreflang tags where every language variant references every other variant plus x-default. Output is delivered as ready-to-paste HTML, a sitemap.xml fragment, and a JSON language map. All variants are self-referencing and reciprocal, matching Google's hreflang specification. Who contributed these multilingual skills? + The blog-multilingual orchestrator was adapted from claude-blog-multilingual by Chris Mueller, submitted to the AI Marketing Hub Pro Hub Challenge in March 2026 and scored 85/100 Proficient. The Claude Blog port removes the original curl-pipe-bash installer and credential handling flagged in the audit, integrates the workflow as core skills, and shares the cultural adaptation reference with blog-localize. // RELATED KEEP READING BLOG MULTILINGUAL One-command multilingual orchestrator. Writes, translates, localizes, and emits hreflang in a single invocation. READ SKILL → BLOG TRANSLATE Format-preserving translation. Markdown, YAML, JSON-LD, SVG, code blocks intact across every target language. READ SKILL → BLOG LOCALIZE Cultural adaptation layer. DACH, Francophone, Hispanic, and Japanese profiles plus custom YAML profiles. READ SKILL → BLOG LOCALE-AUDIT Multilingual QA. Completeness gaps, hreflang breakage, meta parity across all language variants. READ SKILL → PUBLISH IN 4 LANGUAGES TODAY. $ claude plugin install github:AgriciDaniel/claude-blog COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## Write SEO Blog Posts with AI (Free) | Claude Blog URL: https://claude-blog.md/blog/ai-blog-writing-with-claude-code Write SEO Blog Posts with AI (Free) | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/Write SEO Blog Posts with AI // BLOG How to Write SEO Blog Posts with AI in 2026 By Daniel Agrici | Published March 25, 2026 · Updated July 23, 2026 | 8 min read | TUTORIAL AI WRITING Blog content used to cost agencies $2,000 to $5,000 per month. Freelance writers charge $100 to $500 per article. For startups and solo founders, that math does not work. Claude Blog changes the equation entirely. It is free, open-source, and runs in your terminal. This guide walks you through installing Claude Blog, generating your first post, understanding the 4-agent pipeline, choosing the right template, and scoring your content for quality. By the end, you will have a complete workflow for producing SEO-optimized blog posts without spending a dollar on software. What is Claude Blog? Claude Blog is an open-source blog writing skill for Claude Code. Version 2.2.0 uses 31 sub-skills, 5 AI agents, and 12 adaptable content templates to create, review, and improve posts from your terminal. There is no browser dashboard, monthly fee, or usage cap in the skill itself. The tool handles the entire content creation pipeline: research, outlining, drafting, SEO optimization, AI citation optimization, image generation, and quality scoring. It outputs standard Markdown or HTML that works with any CMS, including WordPress, Ghost, Hugo, Jekyll, and Astro. You can explore the full skill documentation at /skills/blog-write. Installation (3 steps) Getting Claude Blog running takes under two minutes. You need Python 3.11+, Claude Code CLI, and a terminal. Step 1: Clone the repository git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git Step 2: Run the installer bash claude-blog/install.sh The installer copies the skill files to your Claude Code configuration directory and sets up the required file structure. No additional dependencies to install. Step 3: Verify installation /blog help If you see the command list with all 16 available commands, you are ready to write your first post. Your first post The simplest way to generate a blog post is the /blog write command followed by your topic: /blog write "how to optimize landing pages for conversions" The /blog write command in action Claude Blog will analyze the topic, select the best template, run the 4-agent pipeline, and output a complete, SEO-optimized article. The default output is Markdown, but you can specify HTML with the format flag. The entire process takes 2 to 4 minutes for a 2,000-word article. The output includes a title, meta description, heading structure, internal link suggestions, and a quality score. The 4-agent pipeline Claude Blog does not generate content in a single pass. It uses four specialized AI agents that work sequentially, each refining the output of the previous agent. This multi-agent approach produces significantly better results than a single prompt. The 4-agent pipeline processes content sequentially, from research through final optimization. Agent 1: Research Agent The Research Agent analyzes your topic, identifies the target keyword, researches competing content, and gathers supporting data points. It determines the optimal word count, heading structure, and content angle based on what is currently ranking for the topic. Agent 2: Outline Agent The Outline Agent takes the research output and creates a detailed content outline. It structures the article with H2 and H3 headings, assigns word counts to each section, identifies where to place statistics, examples, and calls to action, and determines the logical flow of the argument. Agent 3: Writing Agent The Writing Agent generates the full article text based on the outline. It writes each section with the specified word count, incorporates the research data, and follows the selected template structure. It also generates the meta title, meta description, and Open Graph tags. Agent 4: Optimization Agent The Optimization Agent reviews purpose fit, source fidelity, factual support, descriptive metadata, logical headings, contextual links, crawlability, and AI readiness. Style signals provide editing context but do not act as AI-authorship tests or blocking thresholds. Choosing a template Claude Blog includes 12 content templates, each designed for a specific content type. The default behavior auto-selects the best template based on your topic, but you can specify one manually: /blog write "topic" --template comparison Here are all 12 templates with brief descriptions: How-to - Step-by-step instructional guides with numbered processes Listicle - Numbered lists of items, tips, tools, or resources Comparison - Side-by-side analysis of two or more products or approaches Case study - Real-world examples with data, process, and results Pillar page - Comprehensive topic hubs linking to related subtopic articles News/trend - Timely analysis of industry developments and emerging trends Opinion/thought leadership - Expert perspectives backed by data and experience Product review - In-depth analysis of a product with pros, cons, and verdict FAQ/Q&A - Question-and-answer format targeting featured snippets Tutorial - Technical walkthroughs with code examples and screenshots Resource roundup - Curated collections of tools, links, or references Data-driven - Research-heavy articles built around statistics and analysis For a detailed breakdown of each template, see the templates guide. Scoring your post After generating a post, run /blog analyze to score it across 5 weighted categories: /blog analyze path/to/your-post.md Category Weight What it measures Content Quality 30% Depth, accuracy, readability, structure SEO 25% Title tags, meta, headings, keywords, links E-E-A-T 15% Experience, expertise, authoritativeness, trust Technical 15% Schema, canonical, OG tags, sitemap inclusion AI Citation 15% Citability, passage structure, GEO readiness Five weighted categories combine into a single 0-100 content quality score. The scoring system outputs a 0-100 score with specific improvement recommendations. Posts scoring above 80 are considered publication-ready. Posts below 60 need significant revision. Optimizing for AI citations Claude Blog includes built-in GEO (Generative Engine Optimization) support through the /blog geo command. This analyzes your content for AI citation readiness across Google AI Overviews, ChatGPT, Perplexity, and Bing Copilot. The GEO optimization focuses on passage-level citability, structural readability, and entity clarity. It identifies passages that are well-structured for AI extraction and flags sections that need restructuring. You can read more about this at /skills/blog-geo. Integration with Claude SEO Claude Blog works seamlessly with Claude SEO, the companion AI SEO audit skill. While Claude Blog handles content creation, Claude SEO handles site-wide technical SEO audits, schema markup generation, and performance analysis. A typical workflow looks like this: Run /seo audit on your site to identify content gaps Use /blog write to generate articles targeting those gaps Score the output with /blog analyze Publish, then re-run /seo audit to verify improvements Both tools are free, open-source, and MIT licensed. Together, they provide a complete content and SEO pipeline that costs nothing. // FAQ FREQUENTLY ASKED QUESTIONS Is Claude Blog really free? - Yes. Claude Blog is MIT licensed with no paid tiers, no premium features, and no usage limits. The entire codebase is open-source on GitHub. You only need Claude Code CLI, which requires an Anthropic API key or a Claude Pro subscription. What do I need to run it? + You need Python 3.11 or higher, Claude Code CLI installed and configured, and a terminal. It works on macOS, Linux, and Windows (via WSL). No additional dependencies or cloud services required. How long does it take to write a post? + A typical 2,000-word blog post takes 2-4 minutes to generate through the 4-agent pipeline. This includes research, outlining, writing, and optimization. Editing and final review by a human adds additional time depending on the topic. Can I use it with WordPress? + Yes. Claude Blog supports WordPress, Ghost, Hugo, Jekyll, Astro, and more. It generates standard HTML or Markdown output that can be pasted or imported into any content management system. Does it replace human writers? + No. Claude Blog handles research, drafting, and optimization. You provide the expertise, unique perspective, and final review. The tool accelerates the writing process but does not replace the human judgment needed for quality content. // RELATED KEEP READING CLAUDE BLOG VS JASPER VS WRITESONIC Free AI blog writer vs $49/mo Jasper and $16/mo Writesonic. Feature comparison, pricing, and output quality. READ ARTICLE → 12 BLOG TEMPLATES GUIDE Complete guide to Claude Blog's 12 content templates. How-to, listicle, case study, comparison, and more. READ ARTICLE → AI CONTENT SCORING EXPLAINED How Claude Blog's 100-point scoring system works. 5 weighted categories, scoring bands, and improvement tips. READ ARTICLE → WRITE YOUR FIRST POST IN 2 MINUTES. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## Claude Blog vs Jasper vs Writesonic (2026) URL: https://claude-blog.md/blog/claude-blog-vs-jasper-writesonic Claude Blog vs Jasper vs Writesonic (2026) Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/Claude Blog vs Jasper vs Writesonic // BLOG Claude Blog vs Jasper vs Writesonic in 2026 By Daniel Agrici | March 25, 2026 | 7 min read | COMPARISON Three AI blog writing tools. Three completely different approaches. One costs $49 per month, one costs $16 per month, and one costs nothing. This comparison breaks down the features, pricing, output quality, and ideal use cases for Claude Blog, Jasper, and Writesonic so you can pick the right tool for your workflow. Feature comparison Here is a side-by-side breakdown of the core features across all three tools: Feature Claude Blog Jasper Writesonic Price $0 (MIT) $49/mo $16/mo Interface Terminal (Claude Code) Browser Content Templates 12 50+ 100+ Quality Scoring 100-point, 5 categories Basic SEO Score AI Citation Optimization Yes (GEO) No SEO Integration Claude SEO ecosystem SurferSEO add-on ($59/mo) Built-in basic Open Source Yes (MIT) No Custom Personas 4-dimension framework Brand voice Tone selection Image Generation Gemini API (free) DALL-E (included) Built-in Multi-Agent Pipeline 4 agents Single pass The biggest difference is not price. It is architecture. Claude Blog uses a 4-agent pipeline where specialized agents handle research, outlining, writing, and optimization sequentially. Jasper and Writesonic both use single-pass generation, which is faster but produces less structured output. Pricing breakdown The cost difference over 12 months is significant: Claude Blog: $0/year. MIT licensed, no limits, no premium tiers. Jasper Creator: $588/year ($49/mo). Pro plan is $1,500/year ($125/mo). Writesonic: $192/year ($16/mo). Higher tiers available for teams. Jasper's SEO add-on (SurferSEO integration) costs an additional $59/mo, bringing the total for Jasper + SEO to $1,296/year. Claude Blog includes SEO analysis through the Claude SEO companion skill at no extra cost. Monthly cost comparison. Claude Blog is free and open source with no usage limits. Output quality Output quality depends on the use case. Here is where each tool performs best: Claude Blog excels at long-form SEO content, technical articles, and structured guides. The 4-agent pipeline produces articles with clean heading hierarchies, proper keyword distribution, and AI citation-ready passage structure. The 100-point scoring system catches issues that single-pass tools miss. Jasper excels at brand-consistent content across large teams. Its brand voice feature maintains tone and terminology across hundreds of articles. For enterprise teams that need consistent messaging, Jasper's centralized brand controls are a genuine advantage. Writesonic excels at speed and volume. It generates short-form content faster than either alternative and offers the widest template selection. For social media posts, ad copy, and quick blog drafts that will be heavily edited, Writesonic gets you a starting point faster. When to choose Claude Blog Claude Blog is the right choice if you match any of these profiles: Developers and technical teams comfortable with terminal-based workflows Open-source advocates who prefer tools they can inspect, modify, and extend Cost-conscious creators who need professional output without recurring fees SEO-focused writers who want integrated scoring and optimization AI-forward marketers who need GEO optimization for AI citation platforms The terminal interface is not a limitation. It is a feature. CLI workflows integrate with version control, automation scripts, and CI/CD pipelines in ways that browser tools cannot. When to choose Jasper Jasper makes sense for non-technical marketing teams that need a browser-based interface with enterprise features. If your organization has 10+ content creators who need brand voice consistency and centralized template management, Jasper's collaboration features justify the cost. The browser interface eliminates the learning curve for non-technical team members. When to choose Writesonic Writesonic is the budget SaaS option for teams that need a browser-based tool but cannot justify Jasper's pricing. At $16/mo, it offers a reasonable set of features for quick content generation. Its template library is the largest of the three, though template quantity does not always translate to output quality. The developer advantage Terminal-based tools offer advantages that browser-based tools cannot replicate: Version control: Every generated article lives in your Git repository with full history Automation: Script content generation into CI/CD pipelines and cron jobs Extensibility: Modify the source code to add custom templates, agents, or scoring criteria Privacy: Content never passes through a third-party SaaS dashboard Reproducibility: Same command, same configuration, same output structure every time For developers who already work in the terminal, Claude Blog fits naturally into existing workflows. For non-developers, the learning curve is real, and Jasper or Writesonic may be more practical choices. The best tool is the one that fits your workflow. Claude Blog wins on cost, transparency, and SEO depth. Jasper wins on enterprise collaboration. Writesonic wins on accessibility and template variety. // FAQ FREQUENTLY ASKED QUESTIONS Is Claude Blog really free compared to Jasper? - Yes. Claude Blog is MIT licensed with zero cost. Jasper costs $49/mo for the Creator plan and $125/mo for the Pro plan. Writesonic starts at $16/mo. Claude Blog has no paid tiers, no usage limits, and no premium features locked behind a paywall. Which tool produces better content? + Output quality depends on the prompt and topic. Claude Blog's 4-agent pipeline and 100-point scoring system often produce more structured, SEO-optimized content. Jasper excels at brand voice consistency. Writesonic is fastest for short-form content. For long-form blog posts with SEO requirements, Claude Blog's multi-agent approach typically produces the most publication-ready output. Can I switch from Jasper to Claude Blog? + Yes. Export your Jasper content as text or HTML and use /blog rewrite to optimize it with Claude Blog's pipeline. The rewrite command preserves your core message while improving SEO structure, heading hierarchy, and AI citation readiness. Does Claude Blog work without coding? + It requires a terminal and Claude Code CLI. If you prefer a browser interface, Jasper or Writesonic may be better fits. Claude Blog is designed for developers and technical content creators who are comfortable with command-line tools. // RELATED KEEP READING WRITE SEO BLOG POSTS WITH AI Step-by-step guide to writing SEO blog posts with Claude Blog. Install, configure, and generate your first post. READ ARTICLE → 12 BLOG TEMPLATES GUIDE Complete guide to Claude Blog's 12 content templates. How-to, listicle, case study, comparison, and more. READ ARTICLE → BLOG SEO CHECKLIST 2026 The complete blog SEO checklist. Title tags, meta descriptions, heading hierarchy, schema markup, and AI citations. READ ARTICLE → TRY THE FREE OPTION RIGHT NOW. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## 12 Blog Templates for Every Content Type URL: https://claude-blog.md/blog/content-templates-guide 12 Blog Templates for Every Content Type Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/12 Blog Templates // BLOG 12 Blog Templates for Every Content Type By Daniel Agrici | Published March 25, 2026 · Updated July 23, 2026 | 6 min read | GUIDE TEMPLATES Every blog post has an ideal structure. A how-to guide needs numbered steps. A comparison needs a feature table. A case study needs a results section. Claude Blog ships with 12 content templates that handle this structural thinking for you, so every article starts with the right framework for its content type. This guide explains each template: what it is, when to use it, typical word count, and the key structural elements that make it work. You can specify any template with the --template flag in the /blog write command. 1. How-to template What it is: Step-by-step instructional guides that walk readers through a specific process from start to finish. When to use it: Any time you are teaching someone to do something. Installation guides, configuration walkthroughs, workflow tutorials, and process documentation all fit this template. Coverage guidance: Match the reader's intent and the evidence required. Do not pad the article to a fixed length. Key structural elements: Introduction with prerequisites, numbered steps with clear headings, code blocks or screenshots for technical topics, troubleshooting section, and a summary of what was accomplished. 2. Listicle template What it is: Numbered lists of items, tips, tools, or resources organized around a central theme. When to use it: Curated collections, "top N" roundups, tips and tricks articles, and resource compilations. Listicles are one of the most shareable content formats and perform well in search. Coverage guidance: Match the reader's intent and the evidence required. Do not pad the article to a fixed length. Key structural elements: Brief introduction, numbered items with H2 or H3 headings, consistent structure per item (description, benefit, example), and a conclusion with recommendations. 3. Comparison template What it is: Side-by-side analysis of two or more products, tools, approaches, or strategies. When to use it: "X vs Y" articles, tool comparisons, methodology evaluations, and decision guides. Comparison content targets high-intent search queries from readers ready to make a choice. Coverage guidance: Match the reader's intent and the evidence required. Do not pad the article to a fixed length. Key structural elements: Feature comparison table, individual sections for each option, pricing breakdown, use case recommendations ("choose X if..."), and a clear verdict or summary. 4. Case study template What it is: Real-world examples documenting a specific situation, the approach taken, and the measurable results achieved. When to use it: Client success stories, project retrospectives, experiment results, and before/after analyses. Case studies build trust through concrete evidence. Coverage guidance: Match the reader's intent and the evidence required. Do not pad the article to a fixed length. Key structural elements: Background/context, challenge description, solution approach, implementation details, results with specific metrics, and key takeaways. 5. Pillar page template What it is: Comprehensive topic hubs that cover a broad subject in depth and link to related subtopic articles. When to use it: Building topical authority around core themes. Pillar pages serve as the central hub in a topic cluster strategy, linking to and from 5-15 related articles. Coverage guidance: Match the reader's intent and the evidence required. Do not pad the article to a fixed length. Key structural elements: Table of contents, broad topic overview, multiple H2 sections covering subtopics, internal links to cluster articles, summary section, and resource list. 6. News/trend template What it is: Timely analysis of industry developments, emerging trends, product launches, and market shifts. When to use it: Breaking news coverage, quarterly trend reports, product announcement analysis, and industry predictions. Time-sensitive content that needs to publish quickly. Coverage guidance: Match the reader's intent and the evidence required. Do not pad the article to a fixed length. Key structural elements: Lead paragraph with the key news, context and background, analysis of implications, expert commentary or data, and a forward-looking conclusion. 7. Opinion/thought leadership template What it is: Expert perspectives and arguments backed by data, experience, and industry knowledge. When to use it: Establishing authority on a topic, challenging conventional wisdom, presenting a unique perspective, or advocating for a specific approach. Strong E-E-A-T content. Coverage guidance: Match the reader's intent and the evidence required. Do not pad the article to a fixed length. Key structural elements: Clear thesis statement, supporting arguments with evidence, counterargument acknowledgment, author credentials, and a call to action or conclusion. 8. Product review template What it is: In-depth analysis of a specific product or service with hands-on evaluation. When to use it: Software reviews, tool evaluations, book reviews, and service assessments. Review content targets "review" and "is X worth it" search queries. Coverage guidance: Match the reader's intent and the evidence required. Do not pad the article to a fixed length. Key structural elements: Product overview, feature breakdown, pros and cons list, pricing analysis, comparison with alternatives, verdict with rating, and recommended use cases. 9. FAQ/Q&A template What it is: Question-and-answer format articles targeting specific queries and featured snippet opportunities. When to use it: Addressing a genuine set of related customer questions or creating support content. Use this format for reader value, not as an AI-citation or rich-result shortcut. Coverage guidance: Match the reader's intent and the evidence required. Do not pad the article to a fixed length. Key structural elements: Genuine questions, complete answers, supporting detail where useful, and optional FAQPage markup only when it matches visible content. 10. Tutorial template What it is: Technical walkthroughs with code examples, configurations, and detailed implementation instructions. When to use it: Developer-focused content, technical documentation, API integration guides, and coding tutorials. More technical than how-to guides. Coverage guidance: Match the reader's intent and the evidence required. Do not pad the article to a fixed length. Key structural elements: Prerequisites section, environment setup, step-by-step code blocks, expected output examples, error handling, and a working demo or repository link. 11. Resource roundup template What it is: Curated collections of tools, links, books, courses, or references organized by category. When to use it: "Best tools for X" articles, resource libraries, reading lists, and link collections. Resource content attracts backlinks and social shares. Coverage guidance: Match the reader's intent and the evidence required. Do not pad the article to a fixed length. Key structural elements: Category headings, consistent entry format (name, description, link, key feature), comparison notes, and selection criteria explanation. 12. Data-driven template What it is: Research-heavy articles built around statistics, surveys, experiments, and data analysis. When to use it: Original research, industry benchmarks, survey results, A/B test reports, and data analysis articles. Data-driven content earns the most citations from AI platforms. Coverage guidance: Match the reader's intent and the evidence required. Do not pad the article to a fixed length. Key structural elements: Methodology section, data tables and charts, key findings highlighted, statistical context, limitations disclosure, and a conclusions section with actionable insights. Template selection summary Template Word Count Best For How-to 1,500-2,500 Teaching processes Listicle 1,200-2,000 Curated collections Comparison 1,500-2,500 X vs Y decisions Case study 1,500-2,000 Proving results Pillar page 3,000-5,000 Topical authority News/trend 800-1,200 Timely analysis Opinion 1,200-2,000 Thought leadership Product review 1,500-2,500 Tool evaluations FAQ/Q&A 1,000-1,800 Featured snippets Tutorial 2,000-3,000 Technical guides Resource roundup 1,500-2,500 Link collections Data-driven 2,000-3,500 Original research Template selection changes the structure, not a required word count. A short complete article can remain short, while complex research can expand as needed. If you do not specify a template, Claude Blog auto-selects the best match based on your topic. For manual selection, use /blog write "topic" --template comparison. // FAQ FREQUENTLY ASKED QUESTIONS Which template should I start with? - Choose the structure that matches the reader's task and the evidence you have. A how-to fits a real process, while comparison, case-study, news, and research formats serve different purposes. Can I customize templates? + Yes. Adjust the section structure, evidence plan, headings, media, and output format to fit the audience and brand. Templates are starting points, not ranking recipes. What is the right word count? + Use the length needed to satisfy the reader's intent and support the claims. Short complete articles should not be padded, and complex research can expand when the evidence requires it. Can I create my own templates? + Yes. Add a template that defines purpose, structure, evidence needs, optional media, and output rules. Do not impose mandatory FAQs or raw word-count targets. // RELATED KEEP READING WRITE SEO BLOG POSTS WITH AI Step-by-step guide to writing SEO blog posts with Claude Blog. Install, configure, and generate your first post. READ ARTICLE → AI CONTENT SCORING EXPLAINED How Claude Blog's 100-point scoring system works. 5 weighted categories, scoring bands, and improvement tips. READ ARTICLE → CLAUDE BLOG VS JASPER VS WRITESONIC Free AI blog writer vs $49/mo Jasper and $16/mo Writesonic. Feature comparison and pricing. READ ARTICLE → TRY ALL 12 TEMPLATES FOR FREE. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## AI Content Scoring - 100 Points | Claude Blog URL: https://claude-blog.md/blog/ai-content-scoring-explained AI Content Scoring - 100 Points | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/AI Content Scoring Explained // BLOG How AI Content Scoring Works By Daniel Agrici | Published March 25, 2026 · Updated July 23, 2026 | 5 min read | GUIDE SCORING Every blog post Claude Blog generates gets a score from 0 to 100. That score is not a single number. It is the weighted sum of 5 categories, each measuring a different dimension of content quality. Understanding how scoring works helps you write better content, whether you use Claude Blog or not. Run /blog analyze on any Markdown or HTML file to get your score. The analysis takes under 10 seconds and outputs category breakdowns, specific issues, and improvement recommendations. You can explore the full skill documentation at /skills/blog-analyze. The 5 scoring categories Each category has a fixed weight that determines its contribution to the final score. Here is the breakdown: Category Weight What it measures Content Quality 30% Depth, accuracy, readability, structure, word count SEO 25% Title tags, meta descriptions, headings, keywords, internal links E-E-A-T 15% Experience, expertise, authoritativeness, trustworthiness Technical 15% Schema markup, canonical URLs, OG tags, sitemap inclusion AI Citation 15% Passage citability, structural readability, GEO readiness The 5 scoring categories and their weights in the 100-point system Content Quality (30%) This is the heaviest category because content quality is the foundation everything else builds on. It evaluates: Word count: Does the article meet the target length for its template type? Heading structure: Clean H1 to H2 to H3 hierarchy with no skipped levels Paragraph length: Short paragraphs (2-4 sentences) for readability Readability score: Flesch-Kincaid grade level appropriate for the target audience Content depth: Presence of data, examples, statistics, and actionable advice Introduction quality: Does the opening paragraph clearly state what the reader will learn? SEO (25%) The SEO category checks whether your content follows on-page SEO best practices: Title tag: Under 60 characters, includes primary keyword Meta description: 120-160 characters, compelling and keyword-relevant H1 tag: Exactly one per page, includes primary keyword Keyword distribution: Natural placement in headings, first paragraph, and body Internal links: At least 2-3 links to related content External links: References to authoritative sources Image alt text: Descriptive alt attributes on all images E-E-A-T (15%) Google's E-E-A-T framework (Experience, Expertise, Authoritativeness, Trustworthiness) is critical for content credibility. The scoring checks: Author attribution: Named author with credentials or bio link Publication date: Visible date with datetime attribute Source citations: Claims backed by data, studies, or documentation Experience claims: Methodology, evidence, and results when the article genuinely reports first-hand work Trust signals: HTTPS, privacy policy link, contact information Technical (15%) Technical quality ensures your content is properly formatted for search engines and social platforms: Schema markup: BlogPosting or Article JSON-LD present and valid Canonical URL: Properly set to prevent duplicate content issues Open Graph tags: Title, description, image, and URL for social sharing Sitemap inclusion: URL present in sitemap.xml Mobile responsiveness: Content renders correctly on mobile viewports AI Citation Readiness Heuristic (15%) This internal editorial category reviews whether useful claims and entities are clear and supported. It is not a calibrated citation probability: Source fidelity: Claims accurately reflect suitable cited sources Factual support: Specific claims have evidence and unsupported testing claims are flagged Purpose fit: The article solves the reader's task without padding or page-factory patterns Structural clarity: Headings, lists, tables, and prose are used when they help the reader Entity clarity: Brand names, product names, and people clearly identified Scoring bands The 0-100 score maps to four quality bands: Score Band Meaning 90-100 Excellent Publication-ready, exceeds standards 80-89 Good Publication-ready, minor improvements possible 60-79 Needs Work Targeted improvements required before publishing 0-59 Poor Significant revision needed Score bands from 0 (rewrite) to 100 (exceptional) Scores describe alignment with the project's internal checks. They do not predict rankings, citations, traffic, or how many editing passes a draft needs. Style signals are editing context Burstiness, phrase patterns, punctuation, and type-token ratio can help identify repetitive or unclear prose. They cannot determine whether a person or model wrote the text, and they cannot block publication by themselves. Prose linting remains a transparent project style rule, not a Google requirement or AI detector. How to improve each category Here are targeted improvements for each scoring category: Content Quality: Add original analysis or evidence that serves the reader's actual task. SEO: Use descriptive titles, logical headings, contextual links, and crawlable primary content. E-E-A-T Signals: Cite suitable sources and document genuine experience claims with methodology and results. Technical: Set a canonical URL and keep metadata and structured data consistent with visible content. AI Readiness: Clarify entities and relationships, support factual claims, and remove low-value padding. // FAQ FREQUENTLY ASKED QUESTIONS What is a good score? - The quality gate remains 70, and 80 or higher indicates strong alignment with the internal editorial checks. The result still needs human review because it is not a ranking prediction or calibrated probability. Does the score detect AI authorship? + No. Style signals such as burstiness, phrase patterns, punctuation, and vocabulary diversity can provide editing context, but they do not prove authorship and cannot block publication by themselves. Can I game the scoring? + Padding, keyword repetition, mandatory FAQs, metadata-only freshness, and fixed passage lengths earn no points by themselves. Improvements need to add usefulness, originality, factual support, source fidelity, or technical consistency. How often should I re-score? + Re-score after material edits or when evidence, intent, or technical conditions change. Changing only an updated date does not add freshness credit. // NEW FROM CLAUDE BLOG HOW TO FACT-CHECK AI-WRITTEN BLOG POSTS A claim-by-claim workflow for checking evidence, assigning verdicts, and rewriting unsupported statements. READ ARTICLE → // RELATED KEEP READING WRITE SEO BLOG POSTS WITH AI Step-by-step guide to writing SEO blog posts with Claude Blog. Install, configure, and generate your first post. READ ARTICLE → 12 BLOG TEMPLATES GUIDE Complete guide to Claude Blog's 12 content templates. How-to, listicle, case study, comparison, and more. READ ARTICLE → DUAL OPTIMIZATION GUIDE How to optimize blog posts for both Google rankings and AI citation platforms simultaneously. READ ARTICLE → SCORE YOUR CONTENT IN 10 SECONDS. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## Blog SEO Checklist for 2026 (Free) | Claude Blog URL: https://claude-blog.md/blog/blog-seo-checklist-2026 Blog SEO Checklist for 2026 (Free) | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/Blog SEO Checklist 2026 // BLOG The Ultimate Blog SEO Checklist for 2026 By Daniel Agrici | Published March 25, 2026 · Updated July 23, 2026 | 6 min read | CHECKLIST SEO This checklist covers every SEO element your blog posts need in 2026. It is organized into four categories: on-page, technical, content, and AI citations. Use it before publishing every article, or run /blog seo-check to automate the entire process. Four categories covering 34 checks for complete blog SEO On-page SEO checklist On-page elements are the foundation of blog SEO. These are the signals that tell search engines what your content is about and how to display it in results. Title tag under 60 characters. Include your primary keyword near the front. Make it compelling enough to earn clicks from the search results page. Meta description between 120-160 characters. Summarize the article's value proposition. Include the primary keyword naturally. This is your sales pitch in search results. Exactly one H1 tag. The H1 should match or closely relate to the title tag. It is the main heading visible on the page. Clean heading hierarchy (H1, H2, H3). No skipped levels. H2s for main sections, H3s for subsections. Each heading should describe the content below it. Primary keyword in the first 100 words. Search engines weight early content. Introduce your topic and keyword in the opening paragraph. 2-3 internal links minimum. Link to related articles on your site. Internal links distribute page authority and help search engines discover content. 1-2 external links to authoritative sources. Linking to reputable sources signals that your content is well-researched and trustworthy. Descriptive URL slug. Use the primary keyword in the URL. Keep it short, lowercase, and hyphen-separated. Avoid numbers and dates in the slug. Image alt text on every image. Describe what the image shows. Include keywords naturally when relevant, but prioritize accuracy for accessibility. Technical SEO checklist Technical elements ensure search engines can find, crawl, and index your content correctly. Canonical URL set. Every blog post needs a <link rel="canonical"> tag pointing to its preferred URL. This prevents duplicate content issues. Open Graph tags present. Include og:title, og:description, og:image, and og:url. These control how your content appears when shared on social media. BlogPosting or Article schema markup. Add JSON-LD structured data with headline, author, datePublished, dateModified, and publisher. This helps search engines understand your content structure. URL included in sitemap.xml. Every published blog post URL should appear in your sitemap. Submit the sitemap to Google Search Console and Bing Webmaster Tools. Page loads under 3 seconds. Use Lighthouse or PageSpeed Insights to check. Optimize images, minimize CSS/JS, and use a CDN for static assets. Mobile responsive layout. Test on mobile viewports. Text should be readable without zooming. Buttons and links should be tappable with adequate spacing. HTTPS enabled. All pages must be served over HTTPS. Mixed content (HTTP resources on HTTPS pages) is a ranking penalty. No broken links. Check all internal and external links. Broken links signal neglect to search engines and frustrate readers. Robots meta tag set to index, follow. Unless you intentionally want to exclude a page from search results, ensure it is indexable. Content quality checklist Content quality is the most heavily weighted factor in blog SEO. Search engines evaluate depth, accuracy, and usefulness. Minimum 1,500 words for standard articles. Thin content rarely ranks. Aim for the word count that thoroughly covers the topic without padding. Author byline with credentials. Named authors with linked bios or social profiles build trust. Anonymous content lacks E-E-A-T signals. Publication date visible. Use a <time datetime="YYYY-MM-DD"> element. Fresh content signals relevance. Undated content suggests neglect. Data and statistics cited. Back claims with specific numbers from named sources. "According to [source], [statistic]" is the pattern search engines trust. Readability appropriate for audience. Use short paragraphs (2-4 sentences). Vary sentence length. Avoid jargon unless your audience expects it. Unique perspective or original data. Content that repeats what every other article says does not rank. Add your experience, your data, or your analysis. Clear introduction and conclusion. The introduction should state what the reader will learn. The conclusion should summarize key takeaways. Actionable advice. Readers should be able to do something after reading your article. Provide steps, commands, templates, or frameworks. AI-readiness checklist AI readiness uses the same durable foundations as search: original useful content, source fidelity, factual support, clear entities, crawlability, and a good reader experience. The checklist does not predict citations. Intent-dependent completeness. Keep short complete answers short and expand only when the subject, evidence, or reader task requires it. Descriptive headings and stable anchors. Use questions only when they fit the section. Make key material immediately accessible without banning useful accordions. Applicable crawler access. Review robots directives for the services you intend to support. Access never guarantees inclusion or citation. Optional discovery context. An /llms.txt file can summarize the site, but it earns no readiness points by itself. Entity clarity. Clearly identify brand names, product names, and people. AI systems need unambiguous entity references to cite correctly. Useful structure. Use tables, lists, and prose according to the data and reader task rather than a formatting quota. Rendered primary content. Test what relevant crawlers receive. JavaScript JSON-LD is acceptable to Google when it reaches the rendered DOM, matches visible content, and validates. Optional FAQ markup. Add an FAQ only when it helps readers, and mark it up only when the schema matches visible content. It earns no Google or AI-readiness points here. The most effective blog SEO strategy in 2026 optimizes for both traditional search engines and AI citation platforms simultaneously. Claude Blog's /blog seo-check command covers both in a single pass. // FAQ FREQUENTLY ASKED QUESTIONS What is the most important SEO element for blogs? - No single element replaces useful content that satisfies the reader's purpose. Use a descriptive title, support factual claims, keep important content crawlable, and make metadata and structured data consistent with the visible page. How often should I update my checklist? + Review the checklist when official guidance, site behavior, audience needs, or evidence changes. Update dates should reflect material editorial changes, not a freshness tactic. Does this checklist cover AI search? + Yes. It covers AI readiness through originality, source fidelity, factual support, clear entities, crawlability, and reader usefulness. There is no fixed passage length, mandatory FAQ, special AI schema, or citation guarantee. Can Claude Blog automate this checklist? + The /blog seo-check command automates repeatable checks and reports findings. Human review remains necessary for purpose fit, evidence quality, material changes, and judgment. // RELATED KEEP READING WRITE SEO BLOG POSTS WITH AI Step-by-step guide to writing SEO blog posts with Claude Blog. Install, configure, and generate your first post. READ ARTICLE → AI CONTENT SCORING EXPLAINED How Claude Blog's 100-point scoring system works. 5 weighted categories, scoring bands, and improvement tips. READ ARTICLE → CLAUDE BLOG VS JASPER VS WRITESONIC Free AI blog writer vs $49/mo Jasper and $16/mo Writesonic. Feature comparison and pricing. READ ARTICLE → AUTOMATE YOUR CHECKLIST FOR FREE. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## Optimize for Google AND AI Citations | Claude Blog URL: https://claude-blog.md/blog/dual-optimization-guide Optimize for Google AND AI Citations | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/Dual Optimization Guide // BLOG How to Optimize for Google AND AI Citations By Daniel Agrici | Published March 25, 2026 · Updated July 23, 2026 | 7 min read | GUIDE GEO SEO In 2026, there are two search systems competing for your readers. Google still processes 8.5 billion searches per day. But AI-powered search platforms, including Google's own AI Overviews, now reach billions of users monthly. Optimizing for one while ignoring the other means losing traffic from the other. Dual optimization is the practice of writing content that ranks in traditional Google search and gets cited by AI platforms at the same time. The good news is that most of the structural qualities that earn AI citations also improve traditional SEO performance. You do not need two separate strategies. You need one strategy that serves both systems. Traditional SEO vs generative engine optimization Before building a dual strategy, it helps to understand what each system values differently: Factor Traditional SEO GEO (AI Citations) Goal Rank in top 10 blue links Get quoted in AI answers Key signal Backlinks + relevance Brand mentions + citability Content format Keyword-optimized pages Self-contained answer passages Authority Domain Rating / PageRank Entity presence across web Technical Crawlability + speed AI crawler access + SSR User intent Click to visit page Read answer without visiting Traditional SEO and GEO share core principles but differ in execution details The practical overlap is substantial: publish original, useful, well-supported content and keep it technically accessible. No fixed passage length, heading style, FAQ, or special AI schema guarantees selection. The 5 pillars of dual optimization These five pillars represent the structural elements that serve both traditional search and AI citation platforms simultaneously. 1. Stable, descriptive headings Use headings that describe each section clearly and create stable anchors. A question is appropriate when the section answers a genuine reader question, but no question-heading quota applies. 2. Immediately useful key material Place important material where readers can reach it without unnecessary interaction or padding. A short complete answer can stay short, while a complex subject can expand as evidence and intent require. Accordions are acceptable when key material remains accessible and the interaction does not hide essential content. 3. Structured data that feeds both Schema markup helps Google understand your content and helps AI systems identify entities, authors, and publication dates. The key schema types for dual optimization: BlogPosting or Article: Headline, author, datePublished, wordCount FAQPage: Optional markup only when a visible FAQ genuinely helps readers BreadcrumbList: Site hierarchy for both Google breadcrumbs and AI context Person: Author entity with sameAs links to social profiles Structured data helps machines understand eligible page details when it matches visible content and current documentation. It is not a ranking signal or an AI-citation shortcut. FAQPage earns no Google or AI-readiness points here. 4. Multi-platform authority Traditional SEO relies heavily on backlinks. AI citation systems weigh brand mentions more heavily, especially on YouTube (correlation: 0.737), Reddit, and Wikipedia. For dual optimization, build authority on both fronts: Backlinks: Still matter for Google rankings. Pursue through content quality and outreach. YouTube presence: Create video content about your topics. YouTube mentions have the strongest correlation with AI citations. Reddit participation: Engage genuinely in relevant communities. Reddit is a top citation source for both ChatGPT and Perplexity. Author profiles: Maintain active LinkedIn and GitHub profiles linked via schema markup. 5. Technical access for all crawlers Review robots directives for the services you intend to support. An /llms.txt file can provide optional context, but neither crawler access nor the file guarantees inclusion or citation. Make primary content available in the rendered output and verify it with the relevant crawler and rendering tools. JavaScript-generated JSON-LD is acceptable to Google when it reaches the rendered DOM, matches visible content, and validates. How Claude Blog handles dual optimization Claude Blog's 4-agent pipeline is designed for dual optimization from the start. The Research Agent identifies both keyword targets (for SEO) and citability opportunities (for GEO). The Writing Agent structures content with question-based headings and self-contained answer blocks. The Optimization Agent checks both traditional SEO compliance and AI citation readiness. Two commands make this concrete: /blog write "topic" # Generates dual-optimized content /blog geo post.md # Analyzes AI citation readiness The /blog geo command, documented at /skills/blog-geo, produces platform-specific scores for Google AI Overviews, ChatGPT, Perplexity, and Bing Copilot. It identifies which passages are citable and which need restructuring. For site-wide SEO analysis, Claude Blog integrates with Claude SEO, which provides technical audits, schema generation, and GEO readiness scoring at the domain level. Practical example Here is how dual optimization changes a single section: Before (SEO only): ## Landing Page Best Practices There are many factors that contribute to a high-converting landing page. These include clear CTAs, social proof, fast load times, and mobile responsiveness. After (dual-optimized): ## How do you optimize a landing page for conversions? A high-converting landing page needs four elements: a clear call-to-action above the fold, social proof within the first scroll, page load time under 2.5 seconds, and a mobile-responsive layout. According to Unbounce's 2025 conversion benchmark report, landing pages with all four elements convert at 11.45% compared to 2.35% for pages missing two or more. The "after" version serves Google (keyword in heading, structured format, external citation), serves AI systems (question heading, self-contained answer, specific data point), and serves the reader (clear, actionable, evidence-backed). // FAQ FREQUENTLY ASKED QUESTIONS What is dual optimization? - It is an editorial approach that combines search fundamentals with AI-readiness review: original useful content, source fidelity, factual support, clear entities, crawlability, and visible-content consistency. It does not promise rankings or citations. Do AI citations affect Google rankings? + Do not treat an AI citation as a direct Google ranking factor. Search and AI features can use different systems, while useful, well-supported, accessible content is a sound foundation for both. Which AI platforms can surface web sources? + Products such as Google AI features, ChatGPT search, Perplexity, and Microsoft Copilot can surface web sources. Their behavior changes, so avoid fixed platform weights or guaranteed optimization recipes. How do I measure AI citation performance? + Use analytics and platform reporting when available to observe actual referrals and citations. The /blog geo score is an internal readiness heuristic, not a per-platform probability or performance measurement. // NEW FROM CLAUDE BLOG CLAUDE BLOG V2.2.0 RELEASE Continue with the verified Google currentness, backlog audit, analyzer fixes, and market-ready Brain in v2.2.0. READ ARTICLE → // RELATED KEEP READING WRITE SEO BLOG POSTS WITH AI Step-by-step guide to writing SEO blog posts with Claude Blog. Install, configure, and generate your first post. READ ARTICLE → BLOG SEO CHECKLIST 2026 The complete blog SEO checklist. Title tags, meta descriptions, heading hierarchy, schema markup, and AI citations. READ ARTICLE → CLAUDE WATERMARKS + SYNTHID Claude now watermarks its text and the EU AI Act made marking mandatory. What it means for marketing workflows. READ ARTICLE → OPTIMIZE FOR BOTH FOR FREE. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## Best Claude Code Skills in 2026 | Claude Blog URL: https://claude-blog.md/blog/best-claude-code-skills-2026 Best Claude Code Skills in 2026 | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/Best Claude Code Skills 2026 // BLOG Best Claude Code Skills in 2026: The Complete Guide By Daniel Agrici | Published March 25, 2026 · Updated July 23, 2026 | 9 min read | LISTICLE AI TOOLS Claude Code launched as a terminal-first AI coding assistant. But the real power is not in the base tool. It is in the community-built skills that extend it into specialized workflows. Blog writing, SEO auditing, image generation, content scoring, schema markup - developers have built open-source skills for all of it. This guide covers the Claude Code skills that actually matter in 2026. Not every skill on GitHub is worth installing. We tested dozens and narrowed it down to the ones that deliver real results, are actively maintained, and cost nothing to use. What are Claude Code skills? A Claude Code skill is a package of prompts, instructions, and scripts that adds specialized capabilities to your Claude Code CLI. Think of them as plugins for your terminal AI assistant. Each skill installs with a single command and adds new slash commands to your session. Skills work by providing Claude with domain-specific context, structured workflows, and output templates. A blog writing skill, for example, includes 5 specialized AI agents, 12 content templates, and a 100-point scoring system. None of that is built into Claude Code itself. The skill teaches it how to write like a content team. The skill ecosystem is entirely open-source. Anyone can build and share skills on GitHub. Installation follows a consistent pattern: clone the repository, run the installer, and start using the new commands. No package managers, no build steps, no dependencies beyond Python and Claude Code. The best Claude Code skills in 2026 We evaluated skills based on five criteria: active maintenance, documentation quality, real-world utility, community adoption, and output quality. Here are the eight that stood out. 1. Claude Blog - AI blog writing Claude Blog is a broad content-creation skill for Claude Code. Version 2.2.0 uses 31 sub-skills, 5 AI agents, and 12 adaptable templates to create, review, and improve blog posts from the terminal. The public repository has more than 1,800 stars and 300 forks as of August 26, 2026. The standout feature is the 4-agent pipeline. Instead of generating content in a single prompt, Claude Blog runs four specialized agents sequentially: Research, Outline, Writing, and Optimization. Each agent refines the output of the previous one. The result is content that reads like it came from a human writing team, not a chatbot. Key capabilities include: Blog writing (/blog write) - Full article generation with the 4-agent pipeline Content scoring (/blog analyze) - 100-point scoring across 5 weighted categories AI image generation (/blog image) - Hero images via Gemini API GEO optimization - AI citation readiness for Google AI Overviews, ChatGPT, and Perplexity Editorial workflow - Calendar planning, content briefs, taxonomy management, and fact-checking Claude Blog is MIT licensed, completely free, and has no usage limits. If you write blog content and use Claude Code, this is the first skill you should install. Read the full tutorial on writing SEO blog posts with Claude Blog. git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh Claude Blog command interface 2. Claude SEO - AI SEO audits Claude SEO is the companion skill to Claude Blog. While Claude Blog handles content creation, Claude SEO handles site-wide technical SEO. It audits page speed, schema markup, meta tags, internal linking, Core Web Vitals, and more. With over 3,000 GitHub stars, it is the most popular Claude Code skill overall. The core workflow is simple: run /seo audit on any URL and get a comprehensive technical report. The audit covers on-page SEO, technical performance, accessibility, schema validation, and mobile responsiveness. Each finding includes a severity level and specific fix instructions. Claude SEO also generates JSON-LD schema markup, analyzes competitor pages, identifies content gaps, and monitors keyword rankings. It is particularly strong at finding issues that other SEO tools miss, like orphaned pages, duplicate meta descriptions, and broken canonical chains. Learn more at claude-seo.md or view the GitHub repository. git clone --depth 1 https://github.com/AgriciDaniel/Claude-SEO.git && bash Claude-SEO/install.sh 3. Claude Banana - AI image generation Claude Banana brings AI image generation directly into your terminal. It connects to the Google Gemini API to generate blog hero images, social media graphics, OG images, and product mockups. The skill includes a 5-component prompt formula that produces consistent, on-brand visuals. The key advantage over browser-based image tools is workflow integration. You can generate a blog post with Claude Blog, then immediately generate its hero image with Claude Banana, all without leaving your terminal. The skill outputs WebP images at configurable resolutions and aspect ratios. Claude Banana supports 1:1, 16:9, and custom aspect ratios with resolution options from 512px up to 2K. It also handles batch generation for creating multiple image variants from a single prompt. The output integrates directly with Claude Blog's image asset structure. View the repository at GitHub. 4. Code review skills Several community-built skills focus on automated code review within Claude Code. These skills analyze pull requests, identify security vulnerabilities, check for common anti-patterns, and suggest performance improvements. The best ones integrate with GitHub's API to post review comments directly on PRs. Code review skills typically add commands like /review pr or /review file. They analyze code diffs, flag potential issues, and provide contextual suggestions. Some include language-specific rulesets for Python, JavaScript, TypeScript, Go, and Rust. The value here is consistency. Human code reviewers catch different things depending on their mood, workload, and familiarity with the codebase. An AI reviewer applies the same standards to every pull request, every time. 5. Testing and QA skills Testing skills generate unit tests, integration tests, and end-to-end tests based on your existing codebase. They analyze function signatures, class hierarchies, and API endpoints to produce tests that cover edge cases humans often miss. The strongest testing skills support multiple frameworks: pytest for Python, Jest for JavaScript, and Go's built-in testing package. They generate test files that follow the conventions of your existing test suite, including naming patterns, fixture usage, and assertion styles. Some testing skills also include mutation testing capabilities, where they intentionally introduce bugs into your code to verify that your test suite catches them. This is one of the most underrated uses of AI in development workflows. 6. Documentation generators Documentation skills analyze your codebase and generate README files, API documentation, changelog entries, and inline code comments. They read function signatures, docstrings, and usage patterns to produce documentation that stays synchronized with the code. The best documentation skills go beyond simple docstring generation. They create structured API references with request/response examples, generate architecture decision records (ADRs), and produce onboarding guides for new team members based on codebase analysis. 7. Database and migration skills Database skills help with schema design, migration generation, and query optimization. They analyze your existing database schema, suggest indexes for slow queries, generate migration files for schema changes, and validate data integrity constraints. These skills are particularly useful for teams that manage complex database schemas across multiple environments. They can generate rollback migrations automatically, detect breaking changes before deployment, and produce seed data scripts for development environments. 8. DevOps and infrastructure skills Infrastructure skills generate Dockerfiles, CI/CD pipeline configurations, Kubernetes manifests, and Terraform modules. They analyze your application structure and produce deployment configurations that follow best practices for security, performance, and cost optimization. The most useful DevOps skills include pipeline debugging capabilities. They can analyze failing CI/CD pipelines, identify the root cause of build failures, and suggest fixes. Some also include cost analysis features that estimate cloud infrastructure costs based on your deployment configuration. How to install Claude Code skills Every Claude Code skill follows the same installation pattern. Here is the general process: Step 1: Clone the repository git clone --depth 1 https://github.com/author/skill-name.git The --depth 1 flag performs a shallow clone, which is faster and uses less disk space. You only need the latest version of the skill files. Step 2: Run the installer bash skill-name/install.sh The installer copies skill files to your Claude Code configuration directory, typically ~/.claude/skills/. It creates the required directory structure and sets up any configuration files. No additional dependencies to install. Step 3: Verify the installation Open a Claude Code session and run the skill's help command. For Claude Blog, that is /blog help. For Claude SEO, it is /seo help. If you see the command list, the skill is ready to use. To update a skill, pull the latest changes from the repository and re-run the installer: cd skill-name && git pull && bash install.sh Choosing the right skill for your workflow Not every skill is relevant to every workflow. Here is a quick decision framework: If you need to... Install this skill Write blog posts or marketing content Claude Blog Audit technical SEO on your website Claude SEO Generate images for blog posts or social media Claude Banana Automate code reviews on pull requests Code review skill Generate tests for untested code Testing skill Keep documentation current with the code Documentation generator Manage database schemas and migrations Database skill Configure CI/CD and infrastructure DevOps skill Start with the skill that addresses your biggest bottleneck. For most content teams, that is Claude Blog. For development teams, it is usually code review or testing. You can always add more skills later since they do not conflict with each other. The Claude ecosystem advantage The real power of Claude Code skills emerges when you combine them. Claude Blog, Claude SEO, and Claude Banana were designed to work together as a complete content pipeline. Here is how the workflow looks in practice: Audit your site with Claude SEO to identify content gaps and technical issues Plan your content with Claude Blog's strategy and calendar commands Write articles with Claude Blog's 4-agent pipeline targeting those gaps Generate images with Claude Banana for hero images and OG graphics Score and optimize with Claude Blog's 100-point analyzer Verify SEO compliance with Claude SEO's post-publish audit This entire workflow runs in your terminal. No browser tabs, no SaaS dashboards, no monthly subscriptions. Every tool is free, open-source, and MIT licensed. Compare that to the alternative: Jasper at $49/month, Writesonic at $16/month, and Ahrefs at $99/month. The Claude ecosystem replaces hundreds of dollars in monthly software costs with zero-cost terminal tools. For a detailed comparison of Claude Blog against paid alternatives, see our Claude Blog vs Jasper vs Writesonic breakdown. The ecosystem continues to grow. New skills appear on GitHub regularly, and existing skills receive frequent updates. The open-source model means anyone can contribute features, fix bugs, or fork a skill to customize it for their specific needs. The terminal is becoming the content studio. Claude Code skills are the creative team. Whether you are a solo founder writing your first blog post or a content team managing dozens of articles per month, there is a Claude Code skill that fits your workflow. Start with one, learn the commands, and expand from there. The entire ecosystem is free to explore. The Claude Code open-source ecosystem - three free skills working together // FAQ FREQUENTLY ASKED QUESTIONS What is a Claude Code skill? - A Claude Code skill is a reusable set of instructions, prompts, and scripts that extend Claude Code with specialized capabilities. Skills are installed by cloning a GitHub repository and running an installer. Once installed, they add slash commands to your Claude Code session, like /blog write or /seo audit. Are Claude Code skills free? + Most Claude Code skills are free and open-source under the MIT license. Claude Blog, Claude SEO, and Claude Banana are all completely free with no paid tiers, no premium features, and no usage limits. You only need Claude Code CLI, which requires an Anthropic API key or a Claude Pro subscription. How do I install a Claude Code skill? + Installation is typically three steps: clone the repository with git clone, run the included install script (usually install.sh), and verify with the skill's help command. The installer copies skill files to your Claude Code configuration directory. No build steps or package managers required. Can I use multiple skills together? + Yes. Claude Code skills are designed to work independently or together. For example, you can use Claude Blog to write a post, Claude SEO to audit your site, and Claude Banana to generate hero images, all within the same Claude Code session. Each skill adds its own slash commands without conflicting with others. What is the difference between Claude Blog and Claude SEO? + Claude Blog focuses on content creation: writing blog posts, scoring content quality, generating images, and managing editorial workflows. Claude SEO focuses on site-wide technical SEO: auditing page speed, schema markup, meta tags, backlinks, and Core Web Vitals. They are complementary tools. Claude Blog creates the content, Claude SEO ensures your site is technically optimized to rank it. // RELATED KEEP READING HOW TO WRITE SEO BLOG POSTS WITH AI Step-by-step guide to writing SEO blog posts with Claude Blog. Install, configure, and generate your first post with 4 AI agents. READ ARTICLE → DUAL OPTIMIZATION GUIDE How to optimize content for both traditional search engines and AI citation systems. SEO meets GEO in one workflow. READ ARTICLE → AI CONTENT SCORING EXPLAINED How Claude Blog's 100-point scoring system works. 5 weighted categories, scoring bands, and improvement tips. READ ARTICLE → WRITE YOUR FIRST POST IN 2 MINUTES. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## AI Content Pipeline with Claude Blog | Guide URL: https://claude-blog.md/blog/ai-content-pipeline-guide AI Content Pipeline with Claude Blog | Guide Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/AI Content Pipeline Guide // BLOG How to Build an AI Content Pipeline with Claude Blog By Daniel Agrici | Published March 25, 2026 · Updated July 23, 2026 | 8 min read | TUTORIAL Most content teams still run their blog workflows manually. A writer researches keywords in one tool, drafts in Google Docs, passes the draft to an editor over email, checks SEO in yet another dashboard, and publishes through the CMS. Every handoff introduces delays, version conflicts, and lost context. The process works, but it does not scale. A terminal-based AI content pipeline eliminates the handoffs. Claude Blog chains 7 specialized commands together so that content flows from strategy to published post in a single session. Every step runs in the same environment, on the same machine, with zero context switching. This guide walks you through building that pipeline from scratch. What is an AI content pipeline? An AI content pipeline is a repeatable sequence of steps that takes a topic from idea to published article using AI-assisted tools. Instead of treating each blog post as a one-off project, the pipeline standardizes every phase: planning, briefing, outlining, writing, scoring, optimizing, and distributing. The benefits are significant: Consistency - Every post follows the same quality standards and SEO requirements Speed - A 2,000-word post moves from idea to publish-ready in under 15 minutes Scalability - The same pipeline handles 1 post per week or 10 posts per day Quality control - Automated scoring catches issues before they reach your audience Reproducibility - Any team member can run the pipeline and get consistent results Claude Blog provides 31 sub-skills that can chain together, with each command's output feeding naturally into the next step. The 7-step pipeline Here is the complete pipeline, from initial strategy to published content. Each step maps to one or more Claude Blog commands. The complete 7-step AI content pipeline, from strategy to publish Step 1: Strategy (/blog strategy) Every good pipeline starts with a plan. The /blog strategy command analyzes your niche, identifies topic clusters, and generates a content calendar aligned with your business goals. /blog strategy "B2B SaaS marketing" --months 3 The strategy output includes topic clusters organized by search intent, suggested publishing frequency, and a prioritized list of articles ranked by keyword opportunity. This becomes your roadmap for the next quarter. Instead of guessing what to write about, you start with data-informed topic selection. Run this step monthly or quarterly. It sets the direction for everything that follows. Step 2: Content briefs (/blog brief) Once you know what to write, the /blog brief command generates a detailed content brief for each article. The brief includes the target keyword, secondary keywords, search intent, target word count, competitor analysis, and content angle. /blog brief "how to reduce SaaS churn with onboarding emails" A good brief saves hours of revision later. It gives the writing step clear constraints and targets, so the AI agents know exactly what to produce. The brief also specifies which content template to use, what questions to answer, and what unique angle to take based on gaps in existing search results. Step 3: Outlines (/blog outline) The /blog outline command takes the brief and generates a SERP-informed heading structure. It analyzes the top-ranking pages for your target keyword and builds an outline that covers the topic comprehensively while identifying content gaps you can fill. /blog outline "how to reduce SaaS churn with onboarding emails" The outline includes H2 and H3 headings, estimated word counts per section, notes on what data or examples to include, and suggestions for internal linking opportunities. This structure becomes the skeleton that the writing agent fills in. Step 4: Writing (/blog write) This is where the 4-agent pipeline does its work. The /blog write command runs four specialized AI agents in sequence: Research Agent, Outline Agent, Writing Agent, and Optimization Agent. /blog write "how to reduce SaaS churn with onboarding emails" --template how-to The 4-agent pipeline writing a blog post Each agent builds on the previous agent's output. The Research Agent gathers data and context. The Outline Agent structures the content. The Writing Agent generates the full article. The Optimization Agent refines it for SEO, readability, and E-E-A-T signals. The writing step typically takes 2-4 minutes for a 2,000-word article. You can choose from 12 content templates or let the system auto-select the best template for your topic. The output is a complete article with title, meta description, headings, body text, and suggested internal links. Step 5: Scoring (/blog analyze) Before any post leaves the pipeline, it gets scored. The /blog analyze command evaluates content across 5 weighted categories: Category Weight What it measures Content Quality 30% Depth, accuracy, readability, structure SEO 25% Title tags, meta, headings, keywords, links E-E-A-T 15% Experience, expertise, authoritativeness, trust Technical 15% Schema, canonical, OG tags, sitemap inclusion AI Citation 15% Citability, passage structure, GEO readiness /blog analyze path/to/your-post.md Posts scoring 80 or above are considered publication-ready. Posts between 60 and 79 need targeted improvements. Posts below 60 may need a different approach entirely. The scoring output includes specific, actionable recommendations for each category. This step is the quality gate. Nothing moves forward without passing the score threshold you set for your content standards. Step 6: Optimization (/blog rewrite + /blog seo-check + /blog geo) If a post does not hit your target score, three optimization commands help close the gap. /blog rewrite takes the scoring feedback and rewrites specific sections to address weaknesses. It preserves what is working and focuses on the areas that scored lowest. /blog rewrite path/to/your-post.md --focus seo,content /blog seo-check runs a dedicated SEO audit on the post. It checks keyword placement, heading hierarchy, meta tag length, image alt text, internal link count, and other on-page SEO factors. /blog seo-check path/to/your-post.md /blog geo optimizes the content for AI citation. It analyzes passage-level citability, structural readability, and entity clarity across Google AI Overviews, ChatGPT, Perplexity, and Bing Copilot. /blog geo path/to/your-post.md After running these optimization passes, re-score with /blog analyze to verify improvements. Most posts reach 80+ after a single optimization cycle. Step 7: Publishing (/blog schema + /blog repurpose) The final step prepares your content for distribution. /blog schema generates JSON-LD structured data for your post, including BlogPosting, FAQPage, BreadcrumbList, and Person schema. /blog schema path/to/your-post.md /blog repurpose transforms your blog post into content for other platforms: social media threads, email newsletters, LinkedIn articles, and short-form summaries. /blog repurpose path/to/your-post.md --platforms twitter,linkedin,email With schema markup added and distribution content ready, your post is fully prepared for publishing. Paste the HTML into your CMS, add the schema to your page head, and schedule the repurposed content across your channels. Example workflow Here is what building a single post looks like from start to finish. This example creates a how-to guide about email onboarding sequences. # Step 1: Generate a brief /blog brief "how to reduce SaaS churn with onboarding emails" # Step 2: Create a SERP-informed outline /blog outline "how to reduce SaaS churn with onboarding emails" # Step 3: Write the post with the how-to template /blog write "how to reduce SaaS churn with onboarding emails" --template how-to # Step 4: Score the output /blog analyze output/how-to-reduce-saas-churn.md # Step 5: Optimize if needed /blog rewrite output/how-to-reduce-saas-churn.md --focus seo /blog seo-check output/how-to-reduce-saas-churn.md /blog geo output/how-to-reduce-saas-churn.md # Step 6: Re-score to verify /blog analyze output/how-to-reduce-saas-churn.md # Step 7: Generate schema and distribution content /blog schema output/how-to-reduce-saas-churn.md /blog repurpose output/how-to-reduce-saas-churn.md --platforms twitter,linkedin,email Total time for this workflow: roughly 10-15 minutes, including review. The strategy step is not included here because it runs monthly, not per-post. Notice how each command feeds into the next. The brief informs the outline. The outline guides the writing. The writing output gets scored. The score drives optimization. The optimized post gets structured data and distribution content. Every step has a clear input and output. Scaling to multiple posts The real power of a pipeline shows up when you scale. Instead of running the workflow manually for each post, you can batch operations across multiple topics. A typical weekly workflow for a team publishing 5 posts per week looks like this: Monday - Run /blog brief and /blog outline for all 5 topics from the content calendar Tuesday-Wednesday - Run /blog write for each topic, then /blog analyze to score them Thursday - Run optimization commands on posts that scored below 80, then re-score Friday - Run /blog schema and /blog repurpose for all approved posts, then publish This schedule produces 5 fully optimized, schema-marked, distribution-ready blog posts per week. A single person can manage this workflow. With a content team, you can parallelize writing and optimization across multiple people. The key to scaling is the strategy step. When /blog strategy gives you a 3-month content calendar, you never run out of topics. Every Monday, you pull the next 5 topics from the calendar and feed them into the pipeline. Integration with Claude SEO Claude Blog handles content creation. Claude SEO handles site-wide technical SEO. Together, they form a complete content and SEO pipeline. Here is how they work together: Content gaps - Run /seo audit to find pages with thin content or missing topics, then use /blog write to fill those gaps Schema validation - After /blog schema generates structured data, /seo audit can verify it is correctly implemented on the live page Internal linking - Claude SEO identifies orphan pages and link opportunities, while Claude Blog's writing agent automatically suggests relevant internal links Performance tracking - Use /seo audit periodically to measure how your new content is affecting site-wide SEO metrics Both tools are free, open-source, and MIT licensed. You can learn more about getting started with Claude Blog or explore the full writing skill documentation. // FAQ FREQUENTLY ASKED QUESTIONS How long does it take to set up the pipeline? - The initial setup takes about 5 minutes. Clone the repository, run the installer, and you are ready. Configuring your first strategy and content calendar adds another 10-15 minutes. After that, each post flows through the pipeline in 5-10 minutes depending on length and how many optimization passes you run. Can I automate the entire process? + You can automate every step from strategy through scoring, but human review before publishing is strongly recommended. Claude Blog handles the heavy lifting of research, writing, and optimization. A human should review the final output for accuracy, brand voice, and factual claims before hitting publish. What if a post scores below 80? + Run /blog rewrite with the specific improvement areas flagged by /blog analyze. Then run /blog seo-check and /blog geo to address SEO and AI citation issues. Re-score with /blog analyze. Most posts reach 80+ after one optimization pass. Posts below 60 may need a full rewrite with a different template or angle. Do I need Claude SEO too? + No, but they complement each other well. Claude Blog handles content creation and optimization. Claude SEO handles site-wide technical SEO audits, schema validation, and performance analysis. Using both together gives you a complete content and technical SEO pipeline. Can I customize the pipeline for my workflow? + Yes. Every step is modular. You can skip steps, reorder them, or add your own review stages between commands. Some teams run strategy monthly, briefs weekly, and writing daily. Others batch everything. The pipeline adapts to however your team works. // RELATED KEEP READING HOW TO WRITE SEO BLOG POSTS WITH AI Step-by-step guide to writing SEO blog posts with Claude Blog. Install, configure, and generate your first post with 4 AI agents. READ ARTICLE → 12 BLOG TEMPLATES GUIDE Complete guide to Claude Blog's 12 content templates. How-to, listicle, case study, comparison, and more. READ ARTICLE → BLOG SEO CHECKLIST 2026 The complete SEO checklist for blog posts in 2026. On-page, technical, and AI citation optimization in one guide. READ ARTICLE → BUILD YOUR PIPELINE IN 5 MINUTES. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## Free AI Blog Writing Tools in 2026 | Claude Blog URL: https://claude-blog.md/blog/free-ai-blog-writing-tools-2026 Free AI Blog Writing Tools in 2026 | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/Free AI Blog Writing Tools 2026 // BLOG The Best Free AI Blog Writing Tools in 2026 By Daniel Agrici | Published March 25, 2026 · Updated July 23, 2026 | 7 min read | COMPARISON AI blog writing tools used to cost $49 or more per month. Jasper, Writesonic, and similar platforms built their businesses around monthly subscriptions for AI-generated content. In 2026, the landscape looks different. Several free options now exist that can produce quality blog posts without a recurring bill. But "free" does not mean identical. Each tool has different strengths, different limitations, and different trade-offs. Some cap your monthly output. Others require manual prompting for every paragraph. A few give you a full content pipeline at no cost. This guide compares the best free AI blog writing tools available right now, so you can pick the one that fits your workflow. What to look for in a free AI blog writer Before comparing individual tools, it helps to know what separates a useful AI writer from a glorified autocomplete. Here are the five criteria that matter most for blog content: Output quality - Does the tool produce well-structured, readable prose that sounds natural? Can it handle different content types like tutorials, listicles, and comparisons? SEO optimization - Does it generate meta titles, descriptions, heading structures, and keyword placement automatically? Or do you have to handle all of that yourself? Templates - Does it offer pre-built content structures for different article types? Templates save time and produce more consistent results. Scoring and analysis - Can you measure the quality of the output before publishing? A scoring system helps you catch weak spots and improve content systematically. Limitations - What are the actual restrictions? Word limits, feature gates, required subscriptions, or technical prerequisites. Every free tool has at least one. With those criteria in mind, here are the five best free options in 2026. The best free AI blog writing tools 1. Claude Blog - best overall free option Claude Blog is an open-source AI blog writing skill for Claude Code. It runs entirely in your terminal and is licensed under MIT, meaning there are no paid tiers, no premium features, and no usage caps on the tool itself. Claude Blog includes 31 sub-skills, 5 AI agents, and 12 adaptable content templates covering how-to guides, comparisons, case studies, tutorials, research, and more. The tool also includes a 100-point scoring system that evaluates your content across five weighted categories: Content Quality (30%), SEO (25%), E-E-A-T (15%), Technical (15%), and AI Citation (15%). No other free tool offers anything comparable. You get a numerical score and specific improvement recommendations before you publish. On top of that, Claude Blog includes GEO (Generative Engine Optimization), which analyzes your content for citability by AI search engines like Google AI Overviews, ChatGPT, and Perplexity. This is a feature that most paid tools have not even added yet. The limitation is accessibility. Claude Blog is a terminal tool. You need Claude Code CLI installed and configured, which requires either an Anthropic API key or a Claude Pro subscription. If you are comfortable in the terminal, it is the most powerful free option available. If you prefer a browser-based interface, keep reading. git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh 2. Copy.ai - free tier Copy.ai is a browser-based AI writing platform that offers a free tier. The interface is clean and beginner-friendly, with pre-built workflows for blog posts, social media content, and marketing copy. The free plan gives you access to the core writing features and a limited number of words per month. The platform handles formatting and structure through its guided workflows, which makes it easy to get started without any technical knowledge. The limitation is the free tier cap. Copy.ai restricts free users to a small monthly word allowance. Once you hit the limit, you need to wait until the next billing cycle or upgrade to a paid plan. There is no built-in SEO scoring, no content analysis, and no GEO optimization on the free tier. For occasional blog posts, it works. For consistent weekly publishing, you will hit the ceiling quickly. 3. BlogSEO.ai - freemium BlogSEO.ai is built specifically for blog content, which gives it an edge in understanding article structure. The platform generates blog posts with heading hierarchies and basic SEO considerations baked into the output. The free plan lets you generate a limited number of articles and access basic features. The platform focuses on blog-specific workflows rather than general-purpose writing, which means the output tends to be more structured than what you get from generic AI tools. The limitation is that the free plan restricts access to many of the platform's more useful features. Advanced SEO analysis, keyword research integration, and higher output limits are reserved for paid plans. The free tier is enough to evaluate the tool but may not be sufficient for ongoing content production. 4. ChatGPT - with manual prompting ChatGPT is the most widely used AI tool in the world, and its free tier includes access to GPT-4o mini. You can absolutely use it to write blog posts, but it requires manual work that dedicated blog tools handle automatically. To get a good blog post from ChatGPT, you need to prompt it with specific instructions about structure, tone, heading hierarchy, word count, keyword placement, and meta tag generation. Without those instructions, you get generic prose that lacks SEO optimization and consistent formatting. The strengths are real. ChatGPT excels at research, can handle complex topics, and produces natural-sounding text. For writers who know how to prompt effectively, it is a capable tool. The limitation is everything around the writing. There are no blog templates, no content scoring, no SEO analysis, no GEO optimization, and no structured pipeline. Every post is a blank-slate conversation. You also cannot easily reproduce a consistent format across multiple articles without re-entering your prompts each time. 5. Google Gemini - with manual prompting Google Gemini offers a free tier that is competitive with ChatGPT for general-purpose AI writing. Its main advantage is deep integration with Google Search, which means it can pull in current information and cite real sources during the writing process. For research-heavy blog posts, Gemini can save significant time by finding relevant data points, statistics, and references. The output quality is strong, and the model handles long-form content well. The limitation is the same as ChatGPT: no blog-specific features. There are no templates, no scoring, no SEO optimization layer, and no content pipeline. You need to structure everything manually. Gemini also does not offer a way to save or reproduce prompt templates for consistent article formatting across multiple posts. Feature comparison table Tool Price Templates Scoring SEO GEO Open Source Claude Blog $0 (MIT) 12 100-point Yes Copy.ai Freemium Yes No Limited No BlogSEO.ai Freemium Yes Limited Basic No ChatGPT Free tier No Google Gemini Free tier No Annual cost based on lowest paid tier The table makes the gap clear. Claude Blog is the only free tool that offers templates, scoring, SEO optimization, GEO support, and open-source licensing. The trade-off is that it runs in the terminal rather than a browser. Which free tool should you choose? The best tool depends on your situation. Here is a quick decision framework: You want the most features for $0: Claude Blog. No other free tool matches its depth, and the 4-agent pipeline produces higher quality output than single-pass generation. Start with the getting started tutorial. You need a browser-based tool and write fewer than a few posts per month: Copy.ai's free tier is sufficient for light usage and does not require any technical setup. You want blog-specific features but do not need open-source: BlogSEO.ai focuses on blog content and offers a free plan to get started. You already use ChatGPT or Gemini and just want to add blog writing: Both work for blog content with the right prompts. You will not get SEO optimization or scoring, but the raw writing quality is strong. You write content about multiple topics and need research support: Google Gemini's integration with search makes it the best free option for research-heavy articles that need current data. For most serious bloggers, Claude Blog is the clear winner. The combination of templates, multi-agent generation, 100-point scoring, and GEO optimization simply does not exist in any other free tool. The terminal interface is a learning curve, but it takes under two minutes to install and the /blog write command handles the rest. What about paid alternatives? If you have a budget for content tools, the paid landscape offers more options. Jasper starts at $49 per month and provides team collaboration, brand voice training, and a polished browser interface. Writesonic offers plans starting at $16 per month with a similar feature set. The question is whether those features justify the cost. For solo creators and small teams, Claude Blog delivers comparable or better output quality at no cost. The paid tools primarily add convenience features like browser-based editing, team collaboration dashboards, and customer support. For a detailed breakdown of how Claude Blog compares to the leading paid alternatives, read the full Claude Blog vs Jasper vs Writesonic comparison. The bottom line Free AI blog writing tools can support professional workflows when a person still reviews sources, claims, voice, and purpose. Claude Blog provides 31 sub-skills, a multi-agent workflow, and a 100-point internal editorial heuristic for $0. The right choice depends on your workflow. Terminal users should start with Claude Blog. Browser-first users can try Copy.ai or BlogSEO.ai. And everyone can use ChatGPT or Gemini with the right prompts. But if you want the most powerful free option available, it runs in your terminal. It integrates with Claude SEO for site-wide optimization. And it ships with more features than most paid alternatives. // FAQ FREQUENTLY ASKED QUESTIONS Is Claude Blog completely free? - Claude Blog is MIT licensed with no paid tier in the skill itself. Claude Code, optional model APIs, and external services can have separate access or usage costs. Which free AI writer has the best output quality? + There is no universal winner. Output depends on the topic, sources, instructions, model, and human review. Claude Blog differentiates itself with source checks, specialized workflows, and an internal editorial heuristic. Can free tools replace paid ones? + Sometimes. Claude Blog offers 31 sub-skills and 12 adaptable templates, while paid products may provide browser interfaces, team administration, hosted brand systems, or support. Choose based on workflow and total operating cost. Do free AI writers produce search-ready content? + Some tools can generate metadata and structured drafts, but no tool guarantees rankings. Search-ready work still requires useful original content, suitable sources, factual support, crawlability, and human judgment. What is the catch with free AI writing tools? + Free tools trade convenience, quotas, setup effort, collaboration, or hosted support in different ways. Claude Blog requires terminal familiarity and Claude Code access. Check each provider's current terms before choosing. // RELATED KEEP READING CLAUDE BLOG VS JASPER VS WRITESONIC Free AI blog writer vs $49/mo Jasper and $16/mo Writesonic. Feature comparison, pricing, and output quality. READ ARTICLE → HOW TO WRITE SEO BLOG POSTS WITH AI Step-by-step guide to writing SEO blog posts with Claude Blog. Install, configure, and generate your first post. READ ARTICLE → AI CONTENT SCORING EXPLAINED How Claude Blog's 100-point scoring system works. 5 weighted categories, scoring bands, and improvement tips. READ ARTICLE → WRITE YOUR FIRST POST IN 2 MINUTES. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## Get Google SEO Data for Free (API) | Claude Blog URL: https://claude-blog.md/blog/google-seo-data-free-api-guide Get Google SEO Data for Free (API) | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/Get Google SEO Data for Free // BLOG How to Get Google SEO Data for Free By Daniel Agrici | March 28, 2026 | 8 min read | TUTORIAL Ahrefs costs $99 to $999 per month. Semrush starts at $130. Moz Pro at $99. These tools are excellent, but most of the data they surface comes from the same source: Google. And Google gives that data away for free through its APIs. Claude Blog v1.6.5 adds a new /blog google sub-skill that connects directly to 8 Google APIs. You get Core Web Vitals history, Search Console analytics, PageSpeed audits, URL indexation checks, YouTube video search, NLP entity analysis, GA4 traffic data, and keyword ideas. All from your terminal, at roughly $0.0006 per query. This guide shows you how to set it up and what data you can access at each tier. Key Takeaways Google's free APIs provide Core Web Vitals, Search Console, PageSpeed, and indexation data at $0.0006/query Tier 0 (API key only) unlocks 4 commands in 5 minutes with zero cost YouTube mentions have a 0.737 correlation with AI visibility, the strongest signal measured (Ahrefs, 75K brands) Claude Blog v1.6.5 also adds blog-notebooklm, blog-audio, and plugin compliance updates Shared config file works across both Claude Blog and Claude SEO What Google SEO data can you access for free? Claude Blog organizes Google API access into four credential tiers. Each tier unlocks more data as you add credentials. Tier 0 requires nothing more than a free API key from Google Cloud Console. Tier Credentials APIs Available Setup Time 0 API key (free) PageSpeed Insights, CrUX Core Web Vitals, YouTube Data, NLP Entity Analysis 5 min 1 OAuth or Service Account + Search Console, URL Inspection, Indexing API 10 min 2 + GA4 Property ID + GA4 Organic Traffic Reports 5 min 3 + Ads Developer Token + Google Ads Keyword Planner Varies Most bloggers and content teams will get everything they need from Tier 0 and Tier 1. That covers performance monitoring, indexation tracking, and search analytics, all without spending a dollar on subscriptions. The real cost of SEO data The pricing gap between subscription tools and direct API access is significant. Here is what the same data costs across different providers, calculated annually. Annual cost comparison - subscription tools vs. direct Google API access Google's APIs serve the same underlying data that powers these subscription dashboards. PageSpeed Insights runs the same Lighthouse audit. CrUX provides the same Chrome User Experience Report field data. Search Console is Search Console. The difference is that you are accessing it directly instead of through a third-party interface. To be clear, Ahrefs and Semrush offer features that Google APIs do not, like backlink analysis, competitive keyword research, and site crawling. If you need those features, a subscription makes sense. But for Core Web Vitals, search performance, indexation, and content optimization, the APIs give you everything you need. Setting up Google API access Google Cloud Console - API credentials and enabled services Tier 0: Free API key (5 minutes) Create a project in the Google Cloud Console, enable the APIs you need, and generate an API key. The APIs to enable are PageSpeed Insights API v5, Chrome UX Report API, YouTube Data API v3, and Cloud Natural Language API. Then add the key to Claude Blog's shared config file. # Add your API key to the shared config echo '{"api_key": "YOUR_API_KEY"}' > ~/.config/claude-seo/google-api.json # Test with a PageSpeed audit /blog google pagespeed https://your-site.com With just an API key, you unlock four powerful commands: /blog google pagespeed - Full Lighthouse audit with performance, accessibility, best practices, and SEO scores plus CrUX field data /blog google crux-history - 25 weeks of Core Web Vitals trends (LCP, INP, CLS) from real Chrome users /blog google youtube - Search YouTube for relevant videos to embed in your blog posts /blog google nlp - Extract entities and sentiment from any URL or text for E-E-A-T optimization Tier 1: OAuth or service account (10 minutes) For Search Console API data, you need OAuth credentials or a service account. Create OAuth client credentials in Cloud Console, download the JSON file, and point Claude Blog to it. Search Console query performance data in the terminal # Set up OAuth credentials /blog google setup # Query Search Console performance /blog google gsc sc-domain:your-site.com --days 28 # Check if a URL is indexed /blog google inspect https://your-site.com/blog/your-post # Notify Google of a new or updated URL /blog google index https://your-site.com/blog/new-post Tier 1 adds the most operationally useful commands. Search Console analytics show you which queries bring traffic, what positions you rank for, and where click-through rates can improve. URL Inspection tells you if Google has indexed your page, what canonical it selected, and whether there are mobile usability issues. The Indexing API lets you notify Google immediately when you publish or update content. Tier 2 and 3: GA4 and Google Ads Tier 2 adds GA4 Data API organic traffic reports by connecting your Analytics property. Tier 3 unlocks the Google Ads Keyword Planner API for seed keyword ideas and search volume data. Both are optional and designed for teams that already use these Google products. Core Web Vitals monitoring from your terminal One of the most valuable Tier 0 commands is /blog google crux-history, which pulls 25 weeks of Chrome User Experience Report (CrUX) data for any URL or origin. This is the same field data that Google uses for its Core Web Vitals ranking signal. Google's Core Web Vitals thresholds - the same data CrUX API provides The CrUX history command gives you 25 weekly data points for each metric. This is the same data you would see in the Chrome UX Report dashboard or the "Core Web Vitals" section of Search Console, but accessible directly from your terminal. # Check 25-week CWV trends for your domain /blog google crux-history https://your-site.com # Check a specific page /blog google crux-history https://your-site.com/blog/important-post This integrates with Claude Blog's existing /blog seo-check command. After publishing a post, you can run a PageSpeed audit on the live URL to verify that your content meets Google's performance thresholds. The data flows back into the SEO checklist workflow automatically. YouTube embeds and AI visibility Claude Blog v1.6.5 also adds automatic YouTube video embedding to the writing workflow. This is not just a convenience feature. YouTube presence is now the strongest measurable signal for AI visibility. YouTube mentions show the strongest correlation with AI visibility across ChatGPT, AI Mode, and AI Overviews An Ahrefs study of 75,000 brands found that YouTube mentions have a 0.737 correlation with AI visibility, the strongest single signal measured across ChatGPT, Google AI Mode, and AI Overviews. This outperforms branded web mentions (0.664), Reddit presence (0.518), and every other factor tested. Separately, BrightEdge research found that YouTube is cited 200 times more than any other video platform by AI search engines. In February 2025, YouTube citations in AI Overviews increased by 25%, with technical tutorial videos seeing a 40% jump. Claude Blog's /blog write command now automatically discovers and embeds 2-3 relevant YouTube videos during content generation. The embeds use srcdoc lazy loading, which loads a thumbnail and play button instead of the full YouTube iframe. This reduces embed size from roughly 500KB to about 5KB per video, avoiding performance penalties. For AI crawlers like GPTBot, ClaudeBot, and PerplexityBot that do not execute JavaScript, each embed includes a noscript fallback with the video title, channel name, and description. This ensures the video context is visible to AI systems even without iframe rendering. The writing workflow also generates VideoObject JSON-LD schema for each embedded video, giving search engines structured data about your video content. This feature ties directly into the dual optimization strategy, where content is optimized for both Google rankings and AI citation platforms simultaneously. What else is new in v1.6.5 The Google API integration is the headline feature, but v1.6.5 also includes changes from the v1.6.0 release cycle that are worth noting. blog-notebooklm - A new sub-skill that queries Google NotebookLM for source-grounded research. Upload documents to NotebookLM and Claude Blog can pull citation-backed answers during the research phase. This adds a Tier 1 source quality guarantee to your content. blog-audio - Gemini TTS audio narration with 30 voices and 3 modes (summary, full article, two-speaker dialogue). Generates MP3 audio with HTML5 embed code for adding audio versions of your blog posts. Plugin compliance - All 22 SKILL.md frontmatters were cleaned to pass Claude Code's plugin spec validation. Non-standard fields were removed. 44 tests pass across the full test suite. At the v1.6.5 release, Claude Blog included the Google API workflow documented in this guide. The historical v1.6.5 release notes remain on GitHub; the current product is v2.2.0. Getting started Install Claude Blog, add your Google API key, and run your first command in under 10 minutes. # 1. Clone and install git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh # 2. Add your Google API key mkdir -p ~/.config/claude-seo echo '{"api_key": "YOUR_GOOGLE_API_KEY"}' > ~/.config/claude-seo/google-api.json # 3. Run your first PageSpeed audit /blog google pagespeed https://your-site.com # 4. Check Core Web Vitals history /blog google crux-history https://your-site.com For a complete walkthrough of the writing workflow, see How to Write SEO Blog Posts with AI. For the full list of available commands and sub-skills, visit the blog-write skill page. // FAQ FREQUENTLY ASKED QUESTIONS Is Google API access really free? - Yes. PageSpeed Insights API, CrUX API, YouTube Data API, and Search Console API all have generous free tiers that cover normal usage. Claude Blog's /blog google commands stay well within these limits. The only cost is the Claude Code CLI itself, which requires an Anthropic API key or Claude Pro subscription. What data does this replace from Ahrefs or Semrush? + Claude Blog's Google integration provides Core Web Vitals history (25 weeks of CrUX field data), Search Console analytics (clicks, impressions, CTR, position), URL indexation status, PageSpeed Lighthouse audits, NLP entity analysis, and YouTube video search. It does not replace backlink analysis, competitor keyword gaps, or site crawl features, which still require third-party tools. How long does API setup take? + Tier 0 (free API key from Google Cloud Console) takes about 5 minutes. Tier 1 (OAuth or service account for Search Console access) takes about 10 minutes. The shared config file at ~/.config/claude-seo/google-api.json means you set up credentials once for both Claude Blog and Claude SEO. Does YouTube embedding actually improve AI citations? + According to an Ahrefs study of 75,000 brands, YouTube mentions have a 0.737 correlation with AI visibility, the strongest single signal measured. BrightEdge found that YouTube is cited 200 times more than any other video platform by AI search engines. Claude Blog auto-embeds relevant YouTube videos with proper VideoObject schema to maximize this signal. Can I use this with Claude SEO too? + Yes. Claude Blog and Claude SEO share the same Google API credentials stored at ~/.config/claude-seo/google-api.json. Set up your API keys once and both tools can access Search Console, PageSpeed, and CrUX data. Claude Blog uses it for content-level optimization while Claude SEO uses it for site-wide technical audits. // RELATED KEEP READING BLOG SEO CHECKLIST 2026 The complete SEO checklist for blog posts in 2026. On-page, technical, and AI citation optimization in one guide. READ ARTICLE → HOW TO OPTIMIZE FOR GOOGLE AND AI CITATIONS Dual optimization strategy for Google rankings and AI citation platforms. GEO, AEO, and YouTube correlation data. READ ARTICLE → HOW TO BUILD AN AI CONTENT PIPELINE 7-step terminal workflow from strategy to publishing. Chain Claude Blog commands into a repeatable content pipeline. READ ARTICLE → GET YOUR SEO DATA FOR FREE. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## Publish Blog Posts to WordPress with AI | Claude Blog URL: https://claude-blog.md/blog/wp-mcp-ultimate-wordpress-publishing Publish Blog Posts to WordPress with AI | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/Publish Blog Posts to WordPress with AI // BLOG Publish Blog Posts to WordPress with AI (Free Plugin) By Daniel Agrici | March 29, 2026 | 8 min read | TUTORIAL WORDPRESS You write a blog post in the terminal. Claude Blog generates 2,000 words with SEO optimization, images, proper heading hierarchy, and a quality score. Then you spend the next 10 minutes copying the output, pasting it into WordPress, manually uploading images, setting alt text, choosing categories, adding tags, and clicking Publish. That last step is the bottleneck. WP MCP Ultimate eliminates it entirely. WP MCP Ultimate is a free WordPress plugin that connects any MCP-compatible AI to your WordPress site. Install the plugin, paste one config block into Claude Code, and your terminal can now create posts, upload images, set categories, and publish - all without opening the WordPress dashboard. The complete content pipeline becomes: research, write, optimize, publish. One conversation, zero copy-paste. What is WP MCP Ultimate? WP MCP Ultimate is an open-source WordPress plugin that exposes 58 WordPress abilities through the Model Context Protocol (MCP). MCP is a standard that lets AI tools interact with external systems. Instead of manually operating the WordPress dashboard, your AI client sends structured commands through the plugin's REST API. The plugin works with any MCP-compatible AI client: Claude Code, Cursor, Windsurf, and others. For bloggers using Claude Blog, it means the AI that writes your content can also publish it. One free plugin replaces the entire manual publishing workflow. The 58 abilities cover posts, pages, media, categories, tags, users, comments, menus, widgets, plugins, themes, settings, and site health. For content creators, the key abilities are: Create posts - publish or save as draft with full HTML content Upload media from URL - fetch images and upload them with alt text Set featured images - attach a media item as the post thumbnail Manage categories and tags - create, assign, or update taxonomy terms Update existing posts - edit content, change status, modify metadata Patch content - update specific fields without rewriting the full post The Pipeline: Write, Optimize, Publish in One Conversation Before WP MCP Ultimate, the blogging workflow had a gap. Claude Blog handles the writing. Claude SEO handles the optimization - meta descriptions, keyword targeting, schema markup. But publishing was still manual. Now the full pipeline runs in a single terminal session. The complete content pipeline runs in one terminal conversation. Here is what changes when you add WP MCP Ultimate to the workflow: Step Before (manual) After (with WP MCP) Write content /blog write in terminal Upload images Download, open WP Media, drag files, add alt text AI uploads from URL with alt text automatically Set featured image Click Set Featured Image in WP editor AI sets it in the same command Add categories Check boxes in WP sidebar AI assigns or creates categories Add tags Type tags in WP sidebar AI assigns or creates tags Paste content Copy HTML, paste into WP block editor AI creates the post directly Publish Click Publish button AI publishes or saves as draft Time spent 10-15 minutes per post 0 minutes (automated) For a single post, saving 10 minutes is nice. For an agency publishing 20 posts per week, that is over 3 hours saved every week on publishing alone. Watch It in Action This demo shows a real blog post being rewritten with Claude Blog and published directly to a live WordPress site, including image uploads and category assignment. The entire workflow happens in one terminal conversation. BY AGRICI DANIEL WATCH ON YOUTUBE → Setup in 2 Minutes Getting WP MCP Ultimate running requires three steps. No build tools, no complex configuration, no server-side changes beyond installing a standard WordPress plugin. Step 1: Install the plugin Download WP MCP Ultimate from the GitHub repository or search for it in your WordPress plugin directory. Upload the zip file through Plugins > Add New > Upload Plugin, then activate it. Step 2: Generate an API key Go to Settings > WP MCP Ultimate in your WordPress dashboard. Click "Generate API Key." Copy the key. This key authenticates Claude Code (or any MCP client) with your WordPress installation. You can revoke it at any time. Step 3: Add the MCP config to Claude Code Open your Claude Code settings and add this to the MCP servers section: { "mcpServers": { "wordpress": { "url": "https://your-site.com/wp-json/jesuspended/v1/mcp", "headers": { "X-API-Key": "your-api-key-here" } Replace your-site.com with your WordPress domain and your-api-key-here with the key from step 2. That is the entire setup. For a detailed walkthrough with screenshots, see the setup guide PDF in the GitHub releases. Content Refresh at Scale Publishing new posts is only half the story. Most WordPress sites have dozens of old posts that need updating - outdated statistics, stale information, missing images, weak SEO. WP MCP Ultimate makes content refresh a single-conversation workflow. The process works like this: Find the post - ask Claude to list your WordPress posts, filter by category or date range Pull the content - Claude reads the existing post content through the plugin Rewrite it - use /blog rewrite to optimize the content with current data, better structure, and updated statistics Push the update - Claude updates the post in WordPress with the rewritten content, new images, and updated metadata No exporting, no copying between tabs, no manual image re-uploads. If you have 50 posts that need refreshing, you can work through them one conversation at a time. Each update takes minutes instead of the 20-30 minutes of manual editing per post. This pairs well with the AI content pipeline approach. Use /blog audit to identify which posts need attention, then refresh them through the MCP connection. The Full Ecosystem WP MCP Ultimate is one piece of a larger open-source content toolkit. Each tool handles a different stage of the content lifecycle: Claude Blog - writes, scores, and optimizes blog posts (getting started guide) Claude SEO - technical SEO audits, schema markup, keyword targeting, site-wide optimization WP MCP Ultimate - publishes content directly to WordPress with images, categories, and tags Rankenstein - cloud SEO platform for teams All of these tools are built by Daniel Agrici and the open-source community. Claude Blog and Claude SEO are MIT licensed. WP MCP Ultimate is GPL-2.0 licensed (standard for WordPress plugins). Everything is free. The tools work independently, but the real power comes from combining them. Write a post with Claude Blog. Run the SEO check with Claude SEO. Publish it to WordPress through WP MCP Ultimate. Score the live page. Fix any issues. All in the terminal, all in one session. If you want to connect with other content creators using these tools, join the AI Marketing Hub community (free) or the Pro community for advanced workflows and direct support. // FAQ FREQUENTLY ASKED QUESTIONS Is WP MCP Ultimate free? - Yes. WP MCP Ultimate is free and open-source under the GPL-2.0 license. Install it from the WordPress plugin directory or download it from GitHub. There are no paid tiers, no usage limits, and no premium features locked behind a paywall. Does it work with WordPress.com or just self-hosted? + WP MCP Ultimate requires self-hosted WordPress (WordPress.org) because it needs server-side plugin installation and REST API access. WordPress.com hosted plans with plugin support (Business or Commerce tier) should also work, but the plugin is designed and tested for self-hosted WordPress installations. Can I use it without Claude Blog? + Yes. WP MCP Ultimate works with any MCP-compatible AI client, including Claude Code, Cursor, Windsurf, and others. You do not need Claude Blog installed. The plugin simply exposes 58 WordPress abilities via the Model Context Protocol. Claude Blog is optional but recommended for the complete write-to-publish pipeline. Is my WordPress site secure with MCP access? + WP MCP Ultimate uses WordPress REST API authentication with a dedicated API key that you generate from the plugin settings. The key is scoped to the abilities you enable. All communication happens over HTTPS. You control which abilities are active, and you can revoke the API key at any time from your WordPress dashboard. What WordPress abilities does it include? + WP MCP Ultimate includes 58 abilities across posts, pages, media, categories, tags, users, comments, menus, widgets, plugins, themes, settings, and site health. For blogging specifically, you can create and update posts, upload images from URLs with alt text, set featured images, manage categories and tags, and control publication status. // RELATED KEEP READING HOW TO WRITE SEO BLOG POSTS WITH AI Step-by-step guide to writing SEO blog posts with Claude Blog. Install, configure, and generate your first post with 4 AI agents. READ ARTICLE → BUILD AN AI CONTENT PIPELINE Build an automated AI content pipeline from strategy to publishing. 7 steps using Claude Blog's terminal-based workflow. READ ARTICLE → BEST CLAUDE CODE SKILLS IN 2026 The most impactful Claude Code skills in 2026. Blog writing, SEO audits, image generation, and more. Free, open-source tools. READ ARTICLE → WRITE IT. PUBLISH IT. ALL FROM THE TERMINAL. $ git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh COPY GET WP MCP ULTIMATE ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## Obsidian AI Second Brain: The Free Self-Organizing Plugin URL: https://claude-blog.md/blog/claude-obsidian-second-brain Obsidian AI Second Brain: The Free Self-Organizing Plugin Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/Obsidian AI Second Brain // PLUGIN I Turned Obsidian Into a Self-Organizing AI Second Brain - Here's the Free Plugin By Daniel Agrici | May 28, 2026 | 10 min read | PLUGIN SECOND BRAIN Why Most Second Brains Quietly Die The personal knowledge base AI market hit $1.65 billion in 2025 and is growing at 30.3% CAGR toward $6.15 billion by 2030 (Research and Markets, 2026). Yet most people's note vaults are graveyards. Notes go in, links never get made, and six months later you can't find anything. I've been there. Hundreds of notes, almost zero connections, and the low-grade guilt of knowing the system only pays off if I maintain it. The usual advice is "just build a Zettelkasten" or "use bidirectional links." That's the productivity equivalent of telling someone to floss more. Technically correct, practically useless, because the bottleneck was never the method. It's the maintenance. So I built claude-obsidian, a free Claude Code plugin that does the organizing for you. You drop in sources. It reads them, writes linked notes, flags contradictions, and keeps a living index. Your second brain finally maintains itself. A two-minute look at dropping a source in and watching the wiki build itself. Key Takeaways claude-obsidian is a free, MIT-licensed Claude Code plugin built on Andrej Karpathy's LLM Wiki pattern. It reads your sources and writes the linked notes for you. The core is three plain files: a hot cache, an index, and your wiki pages. No database, no embeddings server, plain Markdown you own forever. AI knowledge management saves knowledge workers 30-45% of time on information retrieval (McKinsey, 2025). v1.9 "Compound Vault" adds hybrid retrieval, methodology modes, multi-writer safety, and a thinking framework across 15 skills. What Is an AI Second Brain? An AI second brain is a personal knowledge base where the AI does the filing. You add raw material; the AI reads it, writes summaries, builds links, and keeps the index current. The old "second brain" idea, popular in PKM and Notion circles, still asks you to do the labor. You tag, you link, you maintain. An AI second brain flips that. The model reads the source, extracts the people and concepts, writes the pages, and connects them to everything you already have. Your job shrinks to two steps: feed it good sources and ask good questions. Here's the part people miss. The value of a knowledge base is proportional to its link density, not its note count. A vault with 100 notes and 500 cross-references beats one with 1,000 notes and 50 links. An AI second brain is worth building precisely because the machine is patient enough to make every link, every time. A real claude-obsidian vault: entities, concepts, and sources cross-referenced automatically. What Is Karpathy's LLM Wiki Pattern? Andrej Karpathy, co-founder of OpenAI and former Tesla AI director, shared an approach to personal knowledge management that skips traditional vector search entirely (Karpathy's gist, 2025). Instead of embedding notes into vectors, you keep them as plain Markdown an LLM can read directly. The LLM doesn't just search your notes. It writes and maintains them. The shape is simple. You index source documents into a raw directory. Then you use an LLM to "compile" a wiki: a set of Markdown files with summaries, backlinks, and concept pages that tie everything together. Obsidian is the front end where you browse the raw data, the compiled wiki, and the graph. The LLM owns the busywork. Why does this beat a vector database for personal knowledge? Three reasons, in plain terms: No vector database. No embeddings server to host, no Pinecone bill, no background worker eating RAM. Just Markdown files on your disk. Human-readable output. The wiki is browsable in Obsidian with graph view, links, and full-text search, even with no AI running at all. Compounding intelligence. Each new source gets cross-referenced against everything already there. The 50th source is far more useful than the first. claude-obsidian is my implementation of this pattern. It is not the only one on GitHub, but it is the most complete, with 15 skills, autonomous research, and a hot cache the others don't have. Three plain files at the core: a hot cache, an index, and your wiki pages. No database required. How Do You Build a Second Brain With AI? Building an AI second brain with claude-obsidian is genuinely a three-step loop, and the third step is "do nothing." Once it is set up, you stop filing and hunting for notes entirely. The workflow looks like this: Step 1: Drop a source in Drag a PDF, paste a URL, or dump a meeting transcript into the raw folder. Originals are never modified. They sit there as your immutable record of truth. Step 2: Run "ingest this" The ingest skill reads the source and does the work you'd normally dread. It pulls out named entities like people, tools, and companies. It identifies concepts and frameworks. It creates or updates wiki pages, adds links between related pages, and flags anything that contradicts what you already had. Step 3: There is no step 3 That's it. One command, and a 20-page PDF becomes 8-15 connected wiki pages with proper structure, cross-references, and source tracking. You didn't create a single folder, tag, or link by hand. In my own use, a vault built from a few dozen mixed sources ends up with the large majority of pages linked to at least two others, with zero manual linking on my part. One command takes a raw source to a mesh of linked, sourced pages. What Can claude-obsidian Actually Do? claude-obsidian ships 15 skills as of v1.9, which makes it less of a chat plugin and more of a knowledge operating system. Most tools in that space do one thing. This one covers the full lifecycle. The headline commands, in plain English: /wiki scaffolds the whole vault from a short description. /save files an entire conversation as one clean, linked note. No copy-paste cleanup. /autoresearch takes a topic, reads sources, writes pages, and grows the graph. Up to 12 pages hands-free in a single run. /canvas builds visual reference boards with images and PDFs. /think runs a 10-principle thinking loop for harder decisions, new in v1.9. Under those sit the workhorses: wiki-ingest, wiki-query, wiki-lint, wiki-retrieve, wiki-mode, wiki-cli, and wiki-fold. The point isn't the count. It's that the skills compound. When you ingest a source, the agent doesn't just summarize. It creates entity pages, concept pages, cross-references them against every existing page, and flags conflicts. The 50th source you add weaves 10 new notes into a mesh of 500. Knowledge compounds like interest. Every source you add makes the next answer sharper. Is claude-obsidian a Good Notion Alternative? Yes, if owning your data matters to you. A lot of knowledge-app spend is locked inside proprietary clouds. claude-obsidian goes the other way. Your notes are plain Markdown on your disk. Here's the honest comparison. Notion-style apps are excellent for collaboration and structured databases, but your data lives in their format, on their servers, under their export rules. If they raise prices or shut a feature, your knowledge is hostage. Most AI-memory setups make this worse, not better: they bolt a vector database and a background worker onto your notes, which quietly burns tokens and RAM and adds a second thing that can break. claude-obsidian has neither. The core is three plain files: a hot cache of recent context, an index, and your wiki pages. You can open them in Obsidian, in a text editor, or in Notepad ten years from now. The AI is a tool that reads and writes those files, not a landlord that holds them. That's the whole pitch for the PKM and second-brain crowd: own the data, rent only the intelligence. How Does the Hot Cache Fix Session Amnesia? Every fresh AI conversation starts with total amnesia, and that tax is real. And a chunk of every session gets clawed back re-explaining context to a model that forgot where you left off. claude-obsidian fixes this with a hot cache: a single file holding roughly 500 words of your most recent context. When you start a new conversation, the plugin reads it first and restores the AI's working memory. No recap. No "where were we?" Typically it holds what you were working on, the key decisions and why, which pages changed, and your open next steps. The math is friendly. That's about 500 tokens to read, a rounding error against Claude's context window, and it saves the 2,000 to 3,000 tokens you'd otherwise burn re-establishing context. In my own daily use that's roughly a 4x to 6x return on a tiny token investment, every single session. I walk through giving Claude a memory that never forgets, from an empty vault to a living graph. Before: orphan notes. After: a self-organized, linked graph the AI maintains for you. What Is New in claude-obsidian v1.9 (Compound Vault)? The public release jumped from v1.6 to v1.9.2, and the "Compound Vault" arc is the biggest change since launch. It rests on Anthropic's contextual retrieval research, which cut retrieval failures by up to 49% and up to 67% with reranking (Anthropic, 2024). The short version: your second brain now remembers better and works safely with multiple writers. Four things landed across the arc: Hybrid retrieval. A contextual prefix, BM25 keyword search, and a cosine rerank work together so the wiki pulls the right pages, not just the similar-sounding ones. Methodology modes. Pick how the vault organizes itself: LYT, PARA, Zettelkasten, or Generic. Ingestion and saving route new pages accordingly. Multi-writer safety. Per-file advisory locking means several agents can ingest at once without corrupting your vault. A thinking framework. The new /think skill brings a 10-principle reasoning loop for audits, architecture calls, and ambiguous requests. That's the jump from 14 to 15 skills. I wrote a full walkthrough of the update for anyone who wants the deep dive. Read it here: Claude Obsidian v1.9: the Compound Vault release. How Do You Connect Claude to Obsidian? Connecting Claude to Obsidian with this plugin takes about two minutes. Obsidian's local-file model is exactly why this works: your vault is just a folder of Markdown, so the plugin reads and writes it directly. The retrieval layer behind it leans on research that reduced failures by up to 49% (Anthropic, 2024). The fastest path is two lines: git clone https://github.com/AgriciDaniel/claude-obsidian.git my-wiki bash my-wiki/bin/setup-vault.sh Then open the my-wiki folder in Obsidian and you're done. Prefer the marketplace? Run claude plugin marketplace add AgriciDaniel/claude-obsidian instead. On desktop the plugin uses the Obsidian CLI as its default transport, and it always keeps plain filesystem access as a fallback, so a missing dependency never blocks you. This is the local-first promise in practice. If Claude's API goes down, your wiki still opens and searches in Obsidian. If you switch AI tools next year, your knowledge comes with you because it was never trapped anywhere. For a visual companion to your notes, pair it with claude-canvas to build image and PDF reference boards. The Wiki That Builds Itself The personal knowledge base AI market is climbing from $1.65 billion in 2025 toward $6.15 billion by 2030 (Research and Markets, 2026), and most of that growth will come from tools that do the work humans hate. claude-obsidian is my bet on what that looks like. It's Karpathy's pattern, productized. Fifteen skills covering the full lifecycle from source to maintenance. A hot cache that remembers your context. A Compound Vault that retrieves better and survives multiple writers. And it's free and open source, because knowledge tools should be owned, not rented. Star the repo on GitHub. Read the deep dive: Claude Obsidian v1.9 Compound Vault. Add a visual layer with claude-canvas. // FAQ FREQUENTLY ASKED QUESTIONS What is an AI second brain? - An AI second brain is a personal knowledge base where an AI reads your sources, writes the notes, links them, and keeps everything organized. You add raw material; the AI builds the structure. McKinsey reports knowledge workers save 30-45% of retrieval time with AI knowledge management (McKinsey, 2025). How do you build a second brain with AI? + Drop sources into one folder, then let the AI extract entities, write linked pages, and maintain an index. With claude-obsidian you run one command and a 20-page PDF becomes 8-15 connected notes. No manual folders, tags, or links. The graph compounds with every source you add over time. Is claude-obsidian a good Notion alternative? + Yes, if you want to own your data. Notion stores notes in a cloud-proprietary format you cannot easily move out. claude-obsidian writes plain Markdown that lives on your disk and opens in Obsidian forever. There's no vendor lock-in and no subscription for the plugin, which is MIT licensed and free. How do you connect Claude to Obsidian? + Install claude-obsidian as a Claude Code plugin, then point it at an Obsidian vault folder. The plugin reads and writes the Markdown directly. On desktop it uses the Obsidian CLI by default, with filesystem access as an always-available fallback. Setup takes about two minutes with a two-line clone command. Is it free? + Yes. claude-obsidian is MIT licensed and fully open source on GitHub. You pay only for your AI model usage, such as Claude API tokens, and Obsidian itself is free for personal use. There's no subscription, no vector database to host, and no background worker quietly burning your RAM. // RELATED KEEP READING CLAUDE OBSIDIAN V1.9 COMPOUND VAULT The deep dive on the v1.9 update: hybrid retrieval, methodology modes, and multi-writer safety. READ ARTICLE → BEST CLAUDE CODE SKILLS IN 2026 The most impactful Claude Code skills in 2026. Blog writing, SEO, image generation, and more. READ ARTICLE → HOW TO BUILD AN AI CONTENT PIPELINE Build an automated AI content pipeline from strategy to publishing. 7 steps using Claude Blog. READ ARTICLE → BUILD YOUR SECOND BRAIN IN 2 MINUTES. $ claude plugin marketplace add AgriciDaniel/claude-obsidian COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## Claude Obsidian v1.9 Update: Compound Vault Explained URL: https://claude-blog.md/blog/claude-obsidian-v1-9-compound-vault Claude Obsidian v1.9 Update: Compound Vault Explained Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/Claude Obsidian v1.9 Compound Vault // RELEASE UPDATE Claude Obsidian v1.9: Your AI Second Brain Gets a Compound Memory By Daniel Agrici | May 28, 2026 | 6 min read | RELEASE UPDATE COMPOUND VAULT What Changed in This claude-obsidian Update? The public version of claude-obsidian jumped from v1.6 all the way to v1.9.2, and the headline is the "Compound Vault" arc. The retrieval upgrade behind it is grounded in Anthropic's contextual retrieval research, which cut retrieval failures by up to 49%, and up to 67% once reranking is added (Anthropic, 2024). In short, your second brain now remembers better. Claude Obsidian in two minutes: drop a source, watch the wiki build itself. If you've used claude-obsidian before, here's the quick map of what landed. If you're new, skip down to the primer first, it'll make the rest click. v1.7 Compound Vault: Obsidian CLI as the default transport, hybrid retrieval, and per-file advisory locking for safe multi-agent ingest. v1.8: methodology modes, so the vault organizes itself the way you already think. v1.9: the /think skill, a 10-principle thinking framework, taking the plugin from 14 to 15 skills. v1.9.1 and v1.9.2: audit hardening and prompt-cache hardening, the unglamorous work that keeps it reliable. The Compound Vault adds smarter retrieval and safe multi-writer ingest, all over plain Markdown. Why Does Hybrid Retrieval Matter? Hybrid retrieval is the upgrade you'll feel first. It pairs a contextual prefix with BM25 keyword search and a cosine rerank, and that combination is exactly what catches the near-misses a pure-similarity search hands you. The result: your wiki pulls the right pages, not just the ones that sound similar. Why does this matter for a second brain? Because the failure mode of every knowledge tool is confidently fetching the wrong note. A pure-similarity search will hand you a page that mentions the same words but answers a different question. Layering keyword matching and a rerank on top of context-aware chunks catches the near-misses. You ask, it retrieves, and the answer cites the page it actually came from. Worth being honest here: this is opt-in. The core wiki still runs on plain Markdown with no embeddings server. Hybrid retrieval is there when you want sharper recall on a big vault, not a requirement bolted onto everyone by default. What Are Methodology Modes and Multi-Writer Safety? Two quieter features in the update punch above their weight. Methodology modes and per-file locking both exist for the same reason: a second brain should bend to how you work and survive heavy use. AI knowledge management already saves workers 30-45% of retrieval time (McKinsey, 2025); these features protect that gain at scale. Methodology modes Pick one of four organizing styles: LYT (maps of content plus atomic notes), PARA (projects, areas, resources, archives), Zettelkasten (timestamped, flat, densely linked), or Generic (no opinion). You set the mode once. After that, ingestion, /save, and /autoresearch all route new pages to the right place automatically. The AI files notes the way you already think. Multi-writer safety Per-file advisory locking means multiple agents can ingest into the same vault at once without stepping on each other. Before this, a parallel ingest could corrupt a page mid-write. Now each file write is guarded, so you can run multi-agent research loops and trust the vault on the other side. Per-file locks let several agents ingest at once without corrupting your notes. New Here? What Is claude-obsidian? If this is your first contact with the project, here's the primer. claude-obsidian is a free, MIT-licensed Claude Code plugin that turns Obsidian into a self-organizing AI second brain, built on Andrej Karpathy's LLM Wiki pattern (Karpathy's gist, 2025). You drop in sources; the AI reads them and writes the linked notes. The core is refreshingly boring in the best way: three plain files. A hot cache holding recent context, an index, and your wiki pages, all plain Markdown you own and can open in Obsidian forever. No database, no embeddings server for the core, nothing to migrate or host. Drop any source, ask any question, and the wiki grows richer with every session. For the full guide, with the workflow, the comparison to Notion, and how the hot cache kills session amnesia, read the pillar: how I turned Obsidian into a self-organizing AI second brain. How Do You Get or Update It? Getting the update is a two-line job, and your existing notes are never touched. claude-obsidian stays free and MIT licensed through v1.9, sitting in a market growing at 30.3% CAGR toward $6.15 billion by 2030 (Research and Markets, 2026) while charging nothing for the plugin itself. Fresh install: git clone https://github.com/AgriciDaniel/claude-obsidian.git my-wiki bash my-wiki/bin/setup-vault.sh Already using it? Pull the latest, or re-add via the Claude Code marketplace with claude plugin marketplace add AgriciDaniel/claude-obsidian. Because there's no database, there's nothing to migrate. The update only changes the plugin skills; your Markdown vault carries over untouched. If you want the visual companion, claude-canvas pairs cleanly with it. The Short Version Compound Vault makes your AI second brain remember better, file the way you think, and survive multiple writers, all while staying plain Markdown you own. The retrieval gains are real, and the price is still zero. Star or update the repo on GitHub. New here? Start with the full guide: the self-organizing AI second brain. // FAQ FREQUENTLY ASKED QUESTIONS What is the claude-obsidian Compound Vault update? - Compound Vault is the v1.7 to v1.9 arc. It adds hybrid retrieval, methodology modes, a thinking framework, and per-file locking for multi-writer safety. The retrieval layer draws on Anthropic research that cut retrieval failures up to 49%, and up to 67% with reranking (Anthropic, 2024). What is hybrid retrieval in claude-obsidian? + Hybrid retrieval combines a contextual prefix, BM25 keyword search, and a cosine rerank so your second brain pulls the most relevant pages, not just similar-sounding ones. It is based on Anthropic's contextual retrieval research, designed to surface the right page, not just a similar-sounding one. What are methodology modes? + Methodology modes let you pick how the vault organizes new pages: LYT, PARA, Zettelkasten, or Generic. You set the mode once and ingestion, saving, and autoresearch route pages accordingly. It means the AI files notes the way you already think, instead of forcing one rigid structure on everyone. How do I update to claude-obsidian v1.9? + Pull the latest from the public GitHub repo, or re-add it via the Claude Code marketplace with the marketplace add command. Your existing Markdown vault is untouched; the update only changes the plugin skills. The whole thing stays free and MIT licensed, with no database to migrate at all. Is claude-obsidian still free after the update? + Yes. claude-obsidian remains MIT licensed and fully open source through v1.9. You pay only for your own AI model usage, and Obsidian is free for personal use. The Compound Vault features add no subscription, no hosted vector database, and no background worker burning your RAM. // RELATED KEEP READING OBSIDIAN AI SECOND BRAIN: THE FULL GUIDE The pillar guide to claude-obsidian: workflow, the Notion comparison, and how the hot cache works. READ ARTICLE → BEST CLAUDE CODE SKILLS IN 2026 The most impactful Claude Code skills in 2026. Blog writing, SEO, image generation, and more. READ ARTICLE → HOW TO BUILD AN AI CONTENT PIPELINE Build an automated AI content pipeline from strategy to publishing. 7 steps using Claude Blog. READ ARTICLE → UPDATE YOUR SECOND BRAIN IN 2 MINUTES. $ claude plugin marketplace add AgriciDaniel/claude-obsidian COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## claude-blog v1.6.6-v1.6.8: 29 Accuracy Fixes | Claude Blog URL: https://claude-blog.md/blog/claude-blog-v1-6-8-update claude-blog v1.6.6-v1.6.8: 29 Accuracy Fixes | Claude Blog Skip to main content CLAUDE BLOG HOME SKILLS BLOG HOME SKILLS BLOG GITHUB Home/Blog/claude-blog v1.6.6-v1.6.8 Update // CHANGELOG claude-blog v1.6.6-v1.6.8: 29 Accuracy Fixes, Windows Installer Fix, CI Bumps By Daniel Agrici | April 8, 2026 | 8 min read | CHANGELOG UPDATE Three releases. Twenty-nine corrections. One recurring theme: the data in AI-adjacent documentation degrades faster than the tools themselves. Between v1.6.6 and v1.6.8, claude-blog went through two independent accuracy audits covering 88 items across API specs, content best practices, schema markup, and AI crawler documentation. Here is what changed and why it matters. The Audit Process Starting from v1.6.5, two separate audit passes reviewed every reference file in the skill. Each item was classified: Incorrect: definitively wrong, needs fixing Outdated: was correct, no longer is Confirmed: verified correct Unverifiable: cannot confirm from primary sources The first audit (Areas 1-2) flagged stat attribution problems and unverifiable claims. The second audit (Areas 3-7) covered 88 items and found 7 incorrect, 24 outdated, 40 confirmed, and 17 unverifiable. What follows is a plain-language breakdown of the most important fixes. v1.6.6: Stat Attribution and Unverifiable Claims The first pass focused on statistics referenced in GEO optimization, content rules, and distribution playbook references. The pattern was consistent: stats cited without dates, attributed to secondary sources, or impossible to trace to a primary study. Key changes: Removed multiple statistics that could not be traced to a primary source Added source and date to every retained statistic Corrected attribution errors (stats credited to the wrong organization) Replaced fabricated or misread data with verified alternatives The "46% reading mode" stat was the most egregious case. It appeared in both the GEO optimization and AI crawler guides as a complete misquotation of Kevin Indig's work. The actual Indig findings replaced it: top 10 domains capture 46% of all ChatGPT citations per topic (Growth Memo, March 2026); 44.2% of LLM citations come from the first 30% of page content (Growth Memo, February 2026). v1.6.7: API Accuracy, 29 Corrections Across 10 Files The second audit was more technical, covering the Google API reference files and AI crawler specifications that blog-google and blog-geo rely on. Critical: GA4 Quota Wrong by 8x The GA4 Data API quota was documented as approximately 25,000 tokens per day. The correct figure is 200,000 Core Tokens per day for a standard property (2M for 360 properties), with additional hourly limits of 40,000 per hour per property and 14,000 per hour per project, plus 10 concurrent requests. A quota wrong by 8x is not a rounding error. It changes whether rate limiting is a concern at all for any realistic usage level. CrUX Data Pipeline Status The reference stated CrUX data had migrated out of PageSpeed Insights. As of April 2026, the migration was signaled but not executed. CrUX data still appears in PSI responses. The standalone CrUX API is the recommended long-term path, but the PSI integration is still live. Core Web Vitals vs. Diagnostic Metrics FCP (First Contentful Paint) and TTFB (Time to First Byte) were listed under Core Web Vitals. They are diagnostic metrics. Core Web Vitals are LCP, INP, and CLS only. This distinction matters for how the skill interprets and reports PSI results. Google Ads API Version The reference listed an outdated API version. Updated to v23.1 (February 2026) with a note on the monthly release cadence introduced in January 2026. Any version below v20 is now sunset. AI Crawler Three-Bot Frameworks The crawler guide documented a simplified one-bot-per-company model. The actual architecture is more granular: OpenAI: GPTBot (training) + OAI-SearchBot (search indexing) + ChatGPT-User (retrieval) Anthropic: ClaudeBot (training) + Claude-SearchBot (search indexing) + Claude-User (retrieval). Deprecated: anthropic-ai, claude-web Perplexity: PerplexityBot (indexing) + Perplexity-User (retrieval, may not respect robots.txt) This distinction matters for robots.txt strategy. Blocking training bots is an operator choice with no search consequence. Blocking search and indexing bots means no appearance in AI answers. Additional crawlers added to the guide: Meta-ExternalAgent, Applebot-Extended, Bytespider, Google-Agent (Project Mariner), CCBot, Amazonbot, DuckAssistBot. llms.txt: No Platform Has Confirmed Reading It The reference treated llms.txt as a confirmed best practice. The correct disclaimer: no major AI platform has confirmed reading llms.txt. Google's Gary Illyes stated Google does not support it (July 2025). Semrush testing showed zero crawler visits to llms.txt. Low-cost to implement, but benefits are currently unproven. Schema Fixes (FAQPage, ClaimReview, AggregateRating, ProfilePage) FAQPage disclaimer strengthened: since August 2023, FAQ rich results only show for government and health authority sites. The markup is still valuable for AI citation and Google says not to remove existing markup. ClaimReview marked as deprecated (June 2025 structured data simplification) and removed from active recommendations. AggregateRating: NOT supported on BlogPosting directly. Only on eligible entity types: Product, Recipe, SoftwareApplication, LocalBusiness, Movie, Book. ProfilePage template added for author pages (fully supported since December 2025, helps E-E-A-T signals). Distribution Platform Updates Platform What Changed Reddit API Became paid ($0.24 per 1,000 calls) in June 2023. Free API access is gone. Strategy must rely on organic participation only. LinkedIn 360Brew AI algorithm (late 2024-2025): views down ~50%, engagement down ~25%, external links penalized ~60%. Native content strongly favored. YouTube Shorts Recommendation engine fully decoupled from long-form in late 2025. Shorts no longer boost long-form channel authority. v1.6.8: Windows Installer Fix and CI Bumps Windows PowerShell Installer Fix The one-liner installer used irm | iex, a pattern that throws a ParameterBindingException on Windows PowerShell 5.1. The correct form is iex (irm ...). Both install.ps1 and docs/INSTALLATION.md were updated. Also corrected: the installer summary counted 20 sub-skills (now 22), and the Python version warning incorrectly flagged 3.12+ instead of 3.11+. CI Action Version Bumps actions/checkout: v4 to v6 actions/setup-python: v5 to v6 These supersede Dependabot PRs number 5 and 6. Why This Matters for a Blog Skill claude-blog's reference files are the ground truth its skills cite when advising on SEO strategy, schema markup, and API integration. Wrong data in a reference file propagates to every blog post that references it: incorrect quota warnings, wrong schema recommendations, outdated crawler advice. The audit process, systematically rating every claim as confirmed, outdated, incorrect, or unverifiable, is worth running periodically on any documentation-heavy skill. The rate of factual decay in API documentation, crawler specs, and platform algorithm behavior is high enough that a 6-month-old reference file is worth auditing. Get the Latest Version /plugin marketplace add AgriciDaniel/claude-blog /plugin install claude-blog@agricidaniel-blog-tools Or via direct install: claude plugin install github:AgriciDaniel/claude-blog Full changelog: github.com/AgriciDaniel/claude-blog/releases // FAQ FREQUENTLY ASKED QUESTIONS What is the most critical fix in v1.6.7? - The GA4 Data API quota correction. The previous figure (25,000 tokens per day) was wrong by 8x. The correct limit is 200,000 Core Tokens per day for standard properties. Does the llms.txt fix mean I should not use it? + Not necessarily. llms.txt is low-cost to implement. The correction is that no major AI platform has confirmed reading it, so treat it as speculative rather than confirmed best practice. How do I update to v1.6.8? + Run /plugin marketplace add AgriciDaniel/claude-blog then /plugin install claude-blog@agricidaniel-blog-tools. Or reinstall via claude plugin install github:AgriciDaniel/claude-blog. // RELATED KEEP READING HOW TO WRITE SEO BLOG POSTS WITH AI Step-by-step guide to writing SEO blog posts with Claude Blog. Install, configure, and generate your first post with 4 AI agents. READ ARTICLE → HOW TO GET GOOGLE SEO DATA FOR FREE Access Search Console, Core Web Vitals, and PageSpeed data from your terminal. $0.0006/query vs $500+ per month for Ahrefs. READ ARTICLE → OPTIMIZE FOR GOOGLE AND AI CITATIONS How to optimize blog posts for both Google rankings and AI citation platforms. READ ARTICLE → GET THE LATEST VERSION FREE AND OPEN SOURCE. $ claude plugin install github:AgriciDaniel/claude-blog COPY VIEW ON GITHUB ALL ARTICLES > CLAUDE BLOG V2.2.0 // MIT LICENSE // FREE FOREVER // AGRICIDANIEL GITHUB SKILLS BLOG YOUTUBE LINKEDIN --- ## I Turned Claude Blog Into a Web Skill Vault | Claude Blog URL: https://claude-blog.md/blog/web-skill-mcp-claude-blog I Turned Claude Blog Into a Web Skill Vault | Claude Blog ▥ CLAUDE BLOGFIELD NOTES / SEPTEMBER 2026 HomeBlogSkill vault BUILDING IN PUBLIC · AGENT WORKFLOWS I Turned Claude Blog Into a Web Skill Vault DADaniel Agrici September 19, 2026 One vault. The guidance you need, when you need it. AI-generated conceptual illustration. Listen to the community summary · 1 minute 17 seconds Your browser does not support audio playback. Download the audio summary. Read the audio transcript · This is the original short summary, not a narration of the full article. Hey homies. I turned Claude Blog into a skill vault you can use straight from the web. The idea started with a simple problem. You install skills, collect more skills, update some and forget the others. Eventually, you're maintaining a small museum of instructions. I wanted a different way to use that knowledge: give an agent a website address and a task, let it find the relevant instructions, and get to work. That's what I built with Web Skill MCP. The idea: Keep the skills and their references in a Markdown-first website. Let agents read what they need for the current task, either through the web or a connected MCP client. Individual skill installation is optional; the tools needed to execute the task still matter. watch the experiment In this walkthrough, I give Claude the vault link and ask for a blog outline about which countries have the best tomatoes. Watch it find the outline skill and start researching. Watch on YouTube A website that holds the workflow The foundation is Markdown. The vault holds skill instructions, reference files and guidance on when to use them. You can browse it like a folder of notes, inspect a file, and follow its links to related material. The release we tested on September 16, 2026 contained 32 skills in that explorer, with source versions traceable to the GitHub repository. The vault's source guide explains the import scope and the distinction between repository content and guidance written for the website. That matters because a skill is more than a prompt with a clever title. It can include a workflow, templates, supporting references and scripts. The Agent Skills documentation describes a format built around a SKILL.md file, with supporting material loaded when needed. The folder structure makes those relationships visible. When a writing skill points to a research reference, the agent needs a way to reach that reference. Otherwise, we've moved the first page of the instructions online and left the rest behind. THE ARCHITECTURE One vault. Just what you need. Select a workflow to explore the connection. OutlineWriteReview ⌘blog-outlineSKILL.md ≡ReferencesEvidence & guidance ↗Your briefTopic & audience .mdTHE VAULT ✳YOUR AGENT Outline: load the structure guidance, then plan the article using your brief. Illustrative workflow The agent reads selected guidance. Execution still depends on its tools and your permission. Give your agent a task and the vault URL The easiest way to try it is with an agent that can browse or fetch web pages. Give it the address and a specific outcome. For example: Use https://claude-blog.md/mcp to create a blog draft about how a small team can document its weekly decisions. Write for team leads who want a practical starting point. Read the agent guide, choose the relevant skills, and load the references needed for this task. Research first, outline second, then write and review. Cite sources you actually read. Tell me which checks you couldn't perform. Keep the result as a draft. Ask the agent to find the relevant workflow, read its instructions and apply them. Whether it completes those steps depends on its access and capabilities. You can ask it to name the files it used so you can check whether it followed the references. For a smaller first test, ask for an outline. That gives you a concrete result to inspect before asking the agent to research and write a full article. If your agent supports remote MCP, you can use Connect MCP in the vault. Once connected, the client can search for skills and load them through tools. The compatibility guide covers both routes and their requirements. How this relates to MCP and WebMCP I use Web Skill MCP as the name for this project. It combines a readable website with ways for agents to retrieve the same content. There are three access routes: Route How it works Browser or HTTP The agent reads the Markdown and follows its reference links. Remote MCP A connected client calls tools to search, describe and load skills. Browser WebMCP A supported browser integration calls the tools exposed by the page. Standard MCP connects AI applications to tools and context. Its official SDK documentation describes remote connections alongside other transports. Our endpoint uses that connection to deliver skill guidance. Google's WebMCP introduction describes structured actions exposed by websites for browser agents. In our built-in browser test, the page's WebMCP interface successfully described and loaded the outline skill. That's a result for this browser and this page; it doesn't establish support in every agent. The Markdown route keeps the content usable even when those integrations aren't available. An llms.txt index provides another entry point to relevant files. The agent still needs to find or receive the address and decide what fits the user's task. No local skill install, but real tools still matter Loading a writing skill doesn't magically connect your WordPress account. The vault supplies instructions. Your agent supplies the capabilities to follow them. Live research needs web access, image generation needs an image tool, and publishing needs a configured destination and your permission. Some installed workflows also rely on local scripts. The hosted writing skill declares that its blog_preflight.py validation script isn't bundled. Reading the workflow therefore doesn't mean the agent has run every check that an installed setup could run. This is the distinction I want people to understand before trying it. You can use the hosted guidance without installing each skill. You still need to check whether your agent can complete the particular task. The agent guide and trust guidance keep that use scoped to the user's request. These are instructions for the current job, not permission to change accounts, publish content or override the agent's existing rules. Citations and factual claims still need review. What happened when we tried it We used the live endpoint to retrieve the brief, outline and writing skills, then loaded the supporting research and delivery references. We researched the topic, created an outline and produced a draft. One detail stood out: our broad search returned an ambiguous match. The agent had to inspect the descriptions and choose the right skills. Hosting the library makes it accessible; choosing the right workflow remains part of the work. We also opened the vault in the built-in browser and called its WebMCP tools to describe and load blog-outline. That showed the browser route working for those retrieval actions, alongside the separate remote MCP requests. The experiment notes retain the exact release, tool responses and checks. They remain in the local draft package, together with a browser walkthrough. We haven't measured universal compatibility, token savings or identical output quality against every installed setup. FROM IDEA TO REVIEWED DRAFT 01Research Read sources. Check the claims. 02Outline Choose the structure. Plan the evidence. 03Write Apply the skill. Keep citations clear. 04Review Test the output. State any gaps. Publishing is a separate action after review and authorization. Try one task Open claude-blog.md/mcp, choose a task and ask your agent to use the relevant skill. Start with a brief, an outline or a rewrite. Then inspect the result. Which instructions did it read? Did it follow the references? Did it say when a required tool was missing? That's the experiment I wanted to share with the community: fewer individual installations, a readable home for the workflows, and the same responsibility to check the work. Written by Daniel Agrici, creator of Claude Blog. Adapted from my community audio summary with AI assistance, using the hosted Claude Blog guidance. --- ## Claude Blog project summary URL: https://claude-blog.md/claude-blog.md # Claude Blog Free, open-source AI blog writing skill for Claude Code. ## What It Is Claude Blog is a Claude Code skill ecosystem for creating, scoring, and optimizing useful, source-backed blog content. Version 2.2.0 ships 31 sub-skills, 5 specialized AI agents, 30 commands, 12 adaptable content templates, and 22 reference guides. Its search and AI-readiness guidance follows current Google documentation without promising rankings or citations. ## Install ### Plugin marketplace (recommended, Claude Code 1.0.33+) ``` /plugin marketplace add AgriciDaniel/claude-blog /plugin install claude-blog@agricidaniel-blog ``` ### Manual (Unix, macOS, Linux) ```bash git clone --depth 1 https://github.com/AgriciDaniel/claude-blog.git && bash claude-blog/install.sh ``` ### Windows (PowerShell 5.1+) ```powershell git clone https://github.com/AgriciDaniel/claude-blog.git powershell -ExecutionPolicy Bypass -File claude-blog\install.ps1 ``` Requires Python 3.11+ and Claude Code CLI 1.0.33+. ## The 5-Agent Pipeline When you type `/blog write`, five AI agents process your post in sequence: 1. **Researcher** - Finds current information, verifies sources against quality tiers, and discovers relevant media. 2. **Writer** - Drafts the article using an intent-appropriate template and a logical heading hierarchy. 3. **SEO** - Validates purpose fit, title and metadata, links, schema consistency, crawlability, and social metadata. 4. **Translator** - Produces native-quality translations while preserving markdown, MDX, HTML, frontmatter, schema, and inline SVG. 5. **Reviewer** - Applies the 100-point internal editorial heuristic, verifies factual support, and suggests improvements by severity. ## 12 Content Templates - How-to guide - Listicle - Case study - Comparison - Pillar page - Product review - Thought leadership - Roundup - Tutorial - News analysis - Data research - Knowledge base Templates adapt to search intent and subject complexity. Claude Blog treats word count as intent-dependent completeness, not a Google ranking target. ## 100-Point Scoring System | Category | Points | Focus | |----------|--------|-------| | Content Quality | 30 | Originality, non-commodity value, purpose fit, usefulness | | SEO Optimization | 25 | Descriptive titles, logical headings, links, metadata, crawlability | | E-E-A-T Signals | 15 | Source fidelity, factual support, transparent authorship and methodology | | Technical Elements | 15 | Visible-content/schema consistency, media relevance, mobile and social metadata | | AI Citation Readiness | 15 | Internal readiness heuristic for clear entities, supported claims, and self-contained meaning | Scoring bands: Exceptional (90-100), Strong (80-89), Acceptable (70-79), Below Standard (60-69), Rewrite (under 60). The score is an internal editorial heuristic, not a calibrated probability or a direct Google ranking model. E-E-A-T and all category weights are editorial checks, not direct ranking factors. ## 30 Commands | Command | What it does | |---------|-------------| | `/blog write <topic>` | Write new blog post from scratch | | `/blog rewrite <file>` | Optimize and refresh an existing post | | `/blog analyze <file>` | Quality audit with 0-100 score | | `/blog brief <topic>` | Generate detailed content brief | | `/blog calendar [monthly\|quarterly]` | Editorial calendar | | `/blog strategy <niche>` | Blog strategy with topic ideation | | `/blog outline <topic>` | SERP-informed content outline | | `/blog seo-check <file>` | Post-writing SEO validation | | `/blog schema <file>` | JSON-LD schema generation | | `/blog repurpose <file>` | Repurpose for social, email, YouTube, newsletters | | `/blog geo <file>` | AI citation readiness audit | | `/blog audit [directory]` | Full-site blog health assessment | | `/blog cannibalization [directory]` | Detect keyword overlap across posts | | `/blog factcheck <file>` | Verify statistics against cited sources | | `/blog image [generate\|edit\|setup]` | AI image generation via Gemini | | `/blog persona [create\|list\|apply]` | Writing persona management (NNGroup 4-dimension tone) | | `/blog taxonomy [sync\|audit\|suggest]` | CMS tag/category sync (WordPress, Shopify, Ghost, Strapi, Sanity) | | `/blog notebooklm <question>` | Source-grounded research via NotebookLM | | `/blog audio [generate\|voices\|setup]` | Audio narration via Gemini TTS (30 voices, 80+ languages) | | `/blog google [command] [args]` | Google APIs (PageSpeed, CrUX, GSC, GA4, YouTube, NLP, Keywords) - 13 commands, 4 credential tiers | | `/blog cluster [plan\|execute] <seed>` | Semantic topic-cluster planning + execution (community) | | `/blog multilingual <topic> --languages <codes>` | One-command international publishing (community) | | `/blog translate <file> --to <codes>` | SEO-optimized translation (community) | | `/blog localize <file> --locale <code>` | Cultural deep-adaptation: DACH, FR, ES, JA, custom (community) | | `/blog locale-audit <directory>` | Multilingual QA: completeness, hreflang, meta parity, freshness (community) | | `/blog flow [find\|optimize\|win\|prompts\|sync]` | FLOW framework prompts (30 evidence-led prompts, CC BY 4.0) | | `/blog brand [init\|show\|update]` | Durable BRAND.md + VOICE.md editorial context auto-loaded by all sub-skills | | `/blog chart` | Inline SVG data charts: 7 types, dark-mode, accessible, source-attributed | | `/blog discourse <topic>` | API-free 30-day discourse research (Reddit, X, YouTube, Hacker News) | | `/blog update <file>` | Refresh post with current statistics | ## Major New Features (v1.6.5 to v2.2.0) | Version | Headline | |---------|----------| | **2.2.0** (2026-08-26) | Verified Google currentness, a complete pull-request and issue audit, analyzer fixes, safer installers, adjudicated secret scanning, and a market-ready source-cited Claude Blog Brain. | | **2.1.1** (2026-07-23) | Public distribution normalization, current marketplace instructions, and completed release documentation for the Google-guidance update. | | **2.1.0** (2026-07-23) | Current Google guidance, honest AI-readiness heuristics, canonical and core-update recovery rules, safer installers, auditable Google-update ledger, dependency refreshes, and consistency safeguards. | | **1.9.1** (2026-05-20) | Security hardening + 5-gate Blog Delivery Contract. 95/100 security audit; frontmatter and quality gates enforced pre-publish. | | **1.9.0** (2026-05) | Blog Delivery Contract: a 5-gate pre-publish quality system. The user is never the first reviewer. | | **1.8.0** (2026-05) | `blog-brand` (durable BRAND.md + VOICE.md context), `blog-chart` (inline SVG data charts), and `blog-discourse` (API-free 30-day research) added. | | **1.7.1** (2026-04-27) | Security audit + remediation arc. CRITICAL VULN-001 (API key exposure) closed, OAuth CSRF protection, hash-pinned dependencies across 4 lockfiles, PowerShell 5.1 compatibility fix, frontmatter completeness enforcement. | | **1.7.0** (2026-04-27) | Pro Hub Challenge community release. 5 community skills integrated: `blog-cluster` (Lutfiya Miller, semantic topic clustering), `blog-multilingual` + `blog-translate` + `blog-localize` + `blog-locale-audit` (Chris Mueller, multilingual publishing suite). FLOW framework integration with 30 evidence-led prompts (CC BY 4.0). `blog-translator` agent added. | | **1.6.8** (2026-04-08) | Hotfix: NotebookLM credential handling improvements. | | **1.6.7** (2026-04-08) | Hotfix: Image generation credential management. | | **1.6.5** | Google APIs skill (`blog-google`): 13 commands across PageSpeed, CrUX, GSC, GA4, YouTube, NLP, Keywords. 4-tier credential system (API key, OAuth/SA, GA4, Ads). | ## Industry Detection Claude Blog auto-detects your content type from topic signals: - **Tutorial / how-to**: step-by-step guide template - **Comparison**: feature matrix template (X vs Y) - **News / analysis**: news analysis template with context layer - **Product review**: review template with pros/cons matrix - **Pillar page**: comprehensive authority template Recommendations adapt to your specific content type. ## Companion Skills Claude Blog works alongside, not against, these specialized skills: - [Claude SEO](https://claude-seo.md/) - SEO analysis, schema validation, GEO/AEO audit, Google APIs sharing config with `blog-google` - [Claude Banana](https://github.com/AgriciDaniel/banana-claude) - AI image generation via Gemini ## Community Contributions Five skills landed in v1.7.0 via the AI Marketing Hub Pro Hub Challenge. See [CONTRIBUTORS.md](https://github.com/AgriciDaniel/claude-blog/blob/main/docs/CONTRIBUTORS.md) for full credits. | Contributor | Submission | Integrated as | |------------|-----------|---------------| | **Lutfiya Miller** (Pro Hub Challenge Winner) | Semantic Cluster Engine | `blog-cluster` | | **Chris Mueller** | Multilingual blog suite | `blog-multilingual` + `blog-translate` + `blog-localize` + `blog-locale-audit` + `blog-translator` agent | ## Numbers - 1,800+ GitHub stars - 300+ forks - 14,116 YouTube demo views - 275 YouTube likes - 31 sub-skills - 5 agents (Researcher, Writer, SEO, Translator, Reviewer) - 30 commands - 12 content templates - 22 reference guides - 100-point quality scoring across 5 categories - 80+ languages supported (TTS, translation) - 30 TTS voices (Gemini) - MIT licensed; CC BY 4.0 for FLOW prompts - $0 forever ## Links - Website: https://claude-blog.md - Repository: https://github.com/AgriciDaniel/claude-blog - v2.2.0 release: https://github.com/AgriciDaniel/claude-blog/releases/tag/v2.2.0 - YouTube demo: https://www.youtube.com/watch?v=AeLC4iutG8w - Author: https://github.com/AgriciDaniel - Skool community: https://www.skool.com/ai-marketing-hub ## License MIT for skill code. CC BY 4.0 for bundled FLOW prompts. Free forever. No paid tiers. <!-- AUTOMATION_POSTS_LLMS_FULL:START --> <!-- manifest-set-sha256=ac69a12f25b08404f6d06f4d85320d6dc6b8dc14e30fa978f42c8039c55595ff --> ## How to Fact-Check AI-Written Blog Posts: A Claim-by-Claim Workflow URL: https://claude-blog.md/blog/how-to-fact-check-ai-written-blog-posts Published: 2026-09-07 Modified: 2026-09-07 AI detection and factual verification answer different questions. Run them as two separate checks. A detector may classify writing patterns. It cannot prove that a number, quote, policy statement, or named capability is true. Factual verification starts with the exact claim and the evidence needed to support it. Google says appropriate use of AI is not itself against its guidelines. It also says using automation primarily to manipulate rankings violates spam policies ( Google Search Central , retrieved 2026-09-07). That is a policy point to verify, not a shortcut for deciding whether every sentence in an AI-assisted draft is accurate. Key takeaways Extract claims before you edit style or run an AI detector. Preserve the original sentence, the exact evidence, and a verdict so a reviewer can retrace the decision. Compare source scope, date, conditions, and units before rewriting a claim. Recheck new claims after a substantive rewrite. Build a claim register before you edit Copy every checkable assertion into a register. Include numbers, dates, quotations, policy summaries, named products, and “best” or comparative claims. Keep the exact original wording. Paraphrasing it before checking often hides what the draft actually asserted. Use six verdicts consistently: Supported: the cited evidence supports the original claim at the stated scope and date. Partially supported: the evidence supports only part of the claim, so the final wording must be narrowed to that supported scope. Inferred: the original claim is a clearly labeled conclusion drawn from facts, rather than a source statement. Contradicted: a credible source conflicts with the original claim. Unverified: no adequate evidence was found for the claim. Inaccessible: the cited evidence could not be read or checked. Partial support is a cue to narrow the sentence. Contradicted, unverified, and inaccessible claims are not permission to retain confident copy. Remove, qualify, or escalate the claim until it has a defensible basis. | ID | Exact draft claim | Claim type | Source and retrieval date | Evidence passage or data | Verdict | Final wording | | --- | --- | --- | --- | --- | --- | --- | | C-01 | [copy sentence] | policy, statistic, quote, capability | [URL, YYYY-MM-DD] | [exact support] | supported / partially supported / inferred / contradicted / unverified / inaccessible | [approved rewrite or remove] | Extract claims, then find the strongest evidence Work sentence by sentence through statements a reader could challenge. A source from the organisation responsible for a product, rule, or dataset is usually a better starting point than a roundup repeating it. Record a retrieval date for changeable guidance. If a source is inaccessible, write inaccessible , say why, and do not silently replace it with a search snippet. When two credible sources conflict, do not average their wording. Record both, check whether they cover different product versions, countries, time periods, or definitions, then narrow the claim to what each actually supports. Escalate the conflict if the publication decision depends on it. Compare scope, date, conditions, and units Many bad rewrites begin with a real source used at the wrong scope. Before accepting a claim, compare: Scope: Does the source address this product, audience, or policy? Date: Is the source current enough for the claim being made? Conditions: Are there exclusions, eligibility rules, or methods that the sentence left out? Units: Does a percentage, currency, time period, or measurement match? A price from one plan does not establish every plan's price. A survey result for one country does not describe every market. A result for a one-month period does not establish an annual trend. These are not cosmetic details: they decide whether the original sentence can stay. Worked example: four original claims This example is an editorial exercise using public Google guidance, not a review of a live article. The worked claims were checked against the named Google sources on 2026-09-07. The verdict taxonomy is an editorial classification used for this workflow. ID Exact draft claim Evidence checked Verdict for the original claim Final wording C-01 “Google penalizes all AI-written content.” Google’s AI-content guidance says appropriate AI use is not against its guidelines, while automation used primarily to manipulate rankings violates spam policies. Contradicted “AI use alone does not determine compliance. Google says automation used primarily to manipulate rankings violates its spam policies.” C-02 “This post is comprehensive because it has 2,000 words.” Google’s people-first guidance asks whether content is substantial and complete, but does not set a preferred word count. Contradicted “Assess whether the post gives the reader a complete, useful answer. Do not use word count as proof of comprehensiveness.” C-03 “Adding an author-method note can help readers understand how automation contributed.” Google’s people-first guidance discusses clear sourcing and author background; its AI-content guidance says disclosure can give readers context. Supported Keep the claim with the cited qualification: use process context when it helps readers understand the work. C-04 “Google allows AI-written content.” Google’s AI-content guidance says appropriate AI use is not against its guidelines, while automation used primarily to manipulate rankings violates spam policies. Partially supported “Google says appropriate AI use is not itself against its guidelines. Automation used primarily to manipulate rankings violates its spam policies.” The original numeric word-count claim is intentionally shown as an unsupported rule of thumb. The review does not invent a statistic to rescue it. Rewrite, then review the rewrite Replace only the sentence you can support. Do not turn a source about a narrow case into a broad promise. Mark a conclusion as an inference when it goes beyond the source. Once the edit is made, run the register again: a new example, headline, meta description, or transition may create a fresh claim. Google’s people-first guidance asks whether content has original information, clear sourcing, and a substantial, complete description of its topic ( Google Search Central , retrieved 2026-09-07). That supports a reviewable editorial process. It does not guarantee rankings or prove that a single fact-check produces a good page. A concise prompt for a fact-check tool Use this as an instruction for a tool or reviewer, then verify the returned evidence yourself: Extract every checkable claim from this draft. For each claim, preserve the exact wording, classify it, cite the strongest accessible source with retrieval date, compare scope/date/conditions/units, assign one of supported, partially supported, inferred, contradicted, unverified, or inaccessible, and propose a safe final wording. Do not invent evidence. Flag conflicting or inaccessible sources for review. For a deeper method, see the claim-by-claim Blog Factcheck skill . For a separate discussion of detection and scoring, see the AI content scoring guide . Questions people ask: Is AI detection the same as fact-checking AI-written content? No. A detector may classify writing patterns, while fact-checking tests whether specific numbers, quotes, policy statements, and named capabilities are supported by evidence. Which verdicts should a claim register use? Use supported, partially supported, inferred, contradicted, unverified, and inaccessible consistently. Narrow partially supported claims, then remove, qualify, or escalate claims that lack a defensible basis. What should you do when credible sources conflict? Record both sources, compare their product versions, countries, time periods, and definitions, then narrow the claim or escalate the conflict when publication depends on it. --- ## Claude Blog v2.2.0: Verified Google Guidance and a Market-Ready Brain URL: https://claude-blog.md/blog/claude-blog-v220-release Published: 2026-08-25 Modified: 2026-08-25 Claude Blog v2.2.0 is the evidence release. Every open pull request and issue was treated as a lead, not as proof. Every current Google claim had to point back to a primary source. The complete signed package, changelog, and verification record are available in the v2.2.0 GitHub release . The backlog was reviewed, not copied The release audit records a disposition for every pull request and issue that was open on August 25, 2026. Useful reports were independently reproduced. Unsafe or unrelated patches were rejected even when they contained one valid observation. Pull request 55 was not merged because it duplicated the complete skill tree under a second source path. Pull request 54 remains a product decision because a new cloud image provider changes privacy, billing, output-rights, and maintenance scope. Pull requests 48 and 40 were rejected as submitted because their broader patches conflicted with current distribution or removed established safeguards. Dependency and action-pin reports were accepted only after the current manifests, locks, official releases, and initialization behavior agreed. The full evidence and disposition table is published in the August 2026 repository audit . Google currentness is now executable The repository now checks its reviewed Google ledger against the official Search status data and Search documentation update feed. The check distinguishes three states: current, refresh required, and unavailable. A network outage cannot be reported as proof that guidance is current. The reviewed guidance covers the August 2026 spam update, Search Console reporting incidents, Preferred Sources, review integrity, Google Ads API v25.1, and the continuing rule that useful, original, people-first work matters more than a special AI-search trick. Google's own Search documentation update log and Search status dashboard remain the primary currentness inputs. The analyzer fixes evidence blind spots The analyzer now reads JSON-LD types inside @graph , retains HTML image alt text, recognizes valid non-US government and academic hosts without trusting lookalike domains, and removes code, tables, scripts, styles, and SVG payloads from prose statistics. Turkish articles can use a declared language profile with Ateşman readability while English keeps the Flesch path. These are compatibility improvements, not new ranking promises. The 0 to 100 interface remains an internal editorial heuristic. It does not estimate the probability of a Google ranking or an AI citation. The Brain earned market-ready status The bundled Claude Blog Brain now includes explicit source-review records for all 125 ledger entries, offline verification that rejects stale or incomplete reviews, a fail-closed public projection, and source, memory, provenance, uncertainty, and citation policies. Its executable market-ready audit scored 100 with no warnings or critical failures. The independent strict Brainstein review also scored 100. That score describes the release contract and its deterministic checks. It does not replace human review of advice, rights, visual quality, or live account actions. Verification and installer safety The public candidate passed 344 root tests with one intentional skip and 26 Brain tests. All 125 source-review records passed offline verification. Plugin validation, secret scanning, actionlint, prose hygiene, dependency initialization, repository consistency, live and offline Google currentness, and the public-release validator passed. The matching GitHub Actions run completed successfully. The Windows installer now identifies a reviewed local checkout through $PSScriptRoot . Its isolated install and uninstall smoke installed 32 public skills and five agents, removed the Claude Blog payload, and preserved an unrelated shared marker. Upgrade to v2.2.0 Refresh the public marketplace and install claude-blog@agricidaniel-blog , or clone the repository and check out the signed v2.2.0 tag before running the inspected installer. /plugin marketplace add AgriciDaniel/claude-blog /plugin install claude-blog@agricidaniel-blog Version 2.2.0 does not remove or rename the existing public commands. It makes the evidence, currentness, installer behavior, and release boundary easier to inspect. Questions people ask: Does Claude Blog v2.2.0 remove or rename commands? No. The public command surface remains compatible. The release concentrates on currentness, analyzer correctness, installer safety, dependency evidence, and the bundled Brain. Were all open pull requests merged into v2.2.0? No. Every open pull request and issue was reviewed, but several patches were rejected, held for a product decision, or adopted only after independent verification. Does the 100-point score predict rankings or AI citations? No. It is an internal editorial heuristic for consistent review. It is not a calibrated probability, a direct Google ranking model, or a citation guarantee. How can existing users upgrade to Claude Blog v2.2.0? Refresh the public marketplace and install claude-blog@agricidaniel-blog , or use the signed v2.2.0 tag with the reviewed clone and installer flow. --- ## Claude Blog v2.1.1: Google Guidance, Honest Scoring, Safer Releases URL: https://claude-blog.md/blog/claude-blog-v211-release Published: 2026-07-23 Modified: 2026-07-23 Claude Blog v2.1.1 is a correctness release for anyone who wants useful editorial automation without pretending that a formatting trick can guarantee rankings or AI citations. The public release combines the v2.1.0 Google guidance work with distribution and CI fixes discovered during publication. The complete signed release is available in the v2.1.1 GitHub release . What changed in the scoring model The existing 0 to 100 interface remains compatible, but its meaning is clearer. AI citation scoring is now labeled an internal readiness heuristic, not a calibrated probability. E-E-A-T and every internal weight are editorial heuristics, not direct Google ranking factors. Fixed citation-block and answer-capsule lengths no longer earn points. Question-heading quotas and mandatory FAQ sections no longer earn points. Raw word counts are replaced by intent-dependent completeness. Changing a date without materially changing the article earns no freshness credit. Style signals such as punctuation patterns, burstiness, and phrase lists cannot block delivery by themselves. The model now emphasizes originality, purpose fit, source fidelity, factual support, visible-content and schema consistency, crawlability, media relevance, and reader usefulness. Neutral sourced articles are not instructed to invent first-hand testing. If a page claims testing, the methodology, evidence, and results need to exist. What current Google guidance means in practice The release follows Google's AI features and website guidance : the same foundational SEO practices apply, there is no special AI schema requirement, and useful unique content remains the goal. Area v2.1.1 behavior FAQPage Optional visible-content markup only. It provides no Google score or AI-readiness points. Rendered JSON-LD Accepted when it reaches the rendered DOM, matches visible content, and validates. Canonical fixes Can remain in a pending reevaluation state for up to two weeks after a correct implementation. Core updates Compare after one week, separate Search surfaces, and avoid speculative quick fixes. Discover Checked only when Discover is a declared target, with country relevance and strong large images. Page factories Scaled low-value query variations and fan-out publishing are explicitly prohibited. The canonical workflow follows Google's canonicalization troubleshooting guidance . Request Indexing remains quota-limited and is reserved for important URLs. The update ledger also tracks first-party documentation changes from the Google Search documentation updates page . Technical and release safeguards Technical checks now warn when title, metadata, canonical, schema, or primary content sits beyond Googlebot's first supported two megabytes. Excessive inline base64, CSS, and JavaScript bloat is surfaced without pretending that file size alone is a ranking factor. Back-button warnings require evidence of actual navigation interference, so legitimate History API usage is not flagged. The repository now includes a path-aware consistency graph, a primary-source Google update ledger, installer hash checks, and a public-release validator. The validator catches private repository URLs, the wrong marketplace identity, stale raw links, and private-only instructions before a public push. Dependencies and compatibility Google Gen AI is updated to the tested 2.14.0 release and Patchright to 1.61.2 with regenerated hash locks. CI constructs the audio client and browser driver without API calls or a browser launch. Plugin validation now runs with an exact Claude Code CLI version on pull requests and protected branches. No commands or CLI arguments were removed or renamed. Existing exit behavior, top-level score fields, 0 to 100 totals, and the quality threshold of 70 remain intact. New methodology labels are additive. /plugin marketplace add AgriciDaniel/claude-blog /plugin install claude-blog@agricidaniel-blog What to audit next Existing content should be reviewed for unsupported experience claims, stale mandatory-FAQ advice, fixed passage-length rules, and schema that does not match visible content. Short complete pages should not be padded simply to satisfy a count. Updated dates should represent real editorial changes. v2.1.1 does not promise rankings or citations. It gives writers and reviewers a more honest, inspectable system for producing useful work and documenting why a recommendation exists. Questions people ask: Does Claude Blog v2.1.1 change existing commands? No. Commands, CLI arguments, exit behavior, top-level score fields, 0 to 100 totals, and the quality threshold of 70 remain compatible. Is the AI citation readiness score a probability? No. It is an internal editorial heuristic. It is not a calibrated probability, a Google ranking factor, or a guarantee that an AI system will cite a page. Do articles need an FAQ section or FAQPage schema? No. FAQs are optional when they help readers. Adding an FAQ or FAQPage schema earns no Google or AI-readiness points by itself. How should existing installations upgrade to v2.1.1? Refresh the public marketplace and install claude-blog@agricidaniel-blog , or check out the signed v2.1.1 tag from the public repository. <!-- AUTOMATION_POSTS_LLMS_FULL:END -->