Free, open source blog writer for Claude Code. Researches current sources, writes answer-first posts for Google and AI answers, has an independent reviewer score each one out of 100 and ships only after five delivery gates. 32 skills, 30 /blog commands, 5 agents. Active development at AI-Marketing-Hub/claude-blog; public releases ship here.
Command Reference
Complete reference for 32 skill directories (1 orchestrator + 31 sub-skills);
30 user-facing commands. blog-chart is internal-only, invoked from
blog-write/blog-rewrite.
Every command is invoked through the main orchestrator
(skills/blog/SKILL.md), which routes to the appropriate sub-skill.
The overview table below is the canonical command list. The sections after it are short summaries; the full flow, options and limits of every command live in its sub-skill's
SKILL.md.
Command Overview
/blog <command> [arguments]
| Command | Sub-Skill | Description |
|---|---|---|
write <topic> |
blog-write | Write a new blog post from scratch (v1.9.0: iterates through the 5-gate delivery contract until score 90+ and zero P0, at most 3 failed gate runs per draft) |
rewrite <file> |
blog-rewrite | Optimize an existing blog post (v1.9.0: same delivery contract; rewrites must score at least as high as the original) |
analyze <file-or-url> |
blog-analyze | Audit blog quality with 0-100 score |
brief <topic> |
blog-brief | Generate a detailed content brief |
calendar [monthly|quarterly] [niche] |
blog-calendar | Generate an editorial calendar |
strategy <niche> |
blog-strategy | Blog strategy and topic ideation |
outline <topic> |
blog-outline | SERP-informed outline generation |
seo-check <file> |
blog-seo-check | Post-writing SEO validation |
schema <file> |
blog-schema | Generate JSON-LD schema markup |
repurpose <file> |
blog-repurpose | Repurpose content for other platforms |
geo <file> |
blog-geo | AI citation optimization audit |
audit [directory] |
blog-audit | Full-site blog health assessment |
image [generate|edit|setup] |
blog-image | AI image generation and editing via Gemini |
cannibalization [directory] |
blog-cannibalization | Detect keyword overlap across posts |
factcheck <file> |
blog-factcheck | Verify statistics against cited sources |
persona [create|list|use|show] |
blog-persona | Manage writing personas and voice profiles |
taxonomy [sync|audit|suggest] |
blog-taxonomy | Tag/category CMS management |
notebooklm [ask|discover|library|setup|status|cleanup] [question-or-url] |
blog-notebooklm | Query NotebookLM for source-grounded research |
audio [generate|voices|setup] |
blog-audio | Generate audio narration via Gemini TTS |
google [command] [args] |
blog-google | Google API data: PSI, CrUX, GSC, GA4, NLP, YouTube, Keywords |
cluster [plan|execute] [seed-keyword|cluster-plan.json] |
blog-cluster | Semantic topic-cluster planning + execution (v1.7.0) |
multilingual <topic> --languages <codes> |
blog-multilingual | Write + translate + localize + hreflang (v1.7.0) |
translate <file> --to <codes> |
blog-translate | SEO-optimized translation with format preservation (v1.7.0) |
localize <file> --locale <code> |
blog-localize | Cultural deep-adaptation per locale (v1.7.0) |
locale-audit <directory> |
blog-locale-audit | Multilingual content QA (v1.7.0) |
flow [find|optimize|win|prompts|sync] |
blog-flow | FLOW framework prompts (v1.7.0) |
brand [init|show|update] |
blog-brand | Generate BRAND.md + VOICE.md context auto-loaded by all sub-skills (v1.8.0) |
discourse <topic> |
blog-discourse | API-free last-30-days discourse research; produces DISCOURSE.md (v1.8.0) |
style learn <paths> |
blog-style | Learn author voice profile from existing posts (v1.10.0) |
decay <current-gsc> <previous-gsc> |
blog-decay | Detect content decay: 20%+ QoQ decline from GSC exports (v1.10.0) |
update <file> |
blog-rewrite | Update an existing post with fresh statistics (routes to rewrite) |
Alias: /blog update <file> routes to /blog rewrite <file> for
freshness-focused updates.
/blog brand (v1.8.0)
Generate BRAND.md and VOICE.md at the project root via a short interview.
These files are auto-loaded as fenced untrusted context by every drafting,
review, and audit sub-skill, giving the agent durable brand and voice
guardrails without re-asking on every invocation.
/blog brand init # interactive interview, writes BRAND.md + VOICE.md
/blog brand show # print current BRAND.md and VOICE.md
/blog brand update # re-run a targeted slice of the interview
When you run /blog brand with no subcommand: defaults to show if either
file already exists at the project root, otherwise to init.
The auto-load contract (fence + sanitize + tool-boundary preservation +
provenance) is documented in skills/blog/SKILL.md "Untrusted-Data Contract"
section. See skills/blog-brand/SKILL.md for the full interview script and
output schema.
/blog discourse (v1.8.0)
Research what real practitioners are saying about a topic in the last 30
(or 90) days across Reddit, X / Twitter, YouTube, Hacker News, dev.to,
Medium, GitHub, Stack Overflow, and Substack. API-free: uses WebSearch
with platform-targeted site operators plus recency filters. Produces a
DISCOURSE.md at the project root that downstream /blog write,
/blog brief, and /blog strategy invocations auto-load.
/blog discourse <topic> # default 30-day window
/blog discourse <topic> --days 90 # widen to 90 days
/blog discourse <topic> --input results.json # skip search, use pre-gathered results
/blog discourse <topic> --output brief.md # write to a chosen path
There is no chaining flag. To use the brief downstream, generate DISCOURSE.md
first, then run /blog brief, /blog write, or /blog strategy; the
orchestrator loads DISCOURSE.md through the fenced helper. Flags accepted by
scripts/discourse_research.py: --input, --topic, --days, --output,
--format {markdown,json}, --decomposition.
Workflow phases (full detail in skills/blog-discourse/SKILL.md):
- Topic Pre-Flight (mandatory): runs four keyword-trap checks from
research-quality.md(demographic shopping, numeric trap, overly-literal, generic single-noun). Refusing trap topics saves WebSearch calls. - Decomposition: split into discrete queries (primary entity, counter-perspective, practitioner discourse, tangential entities, time anchor).
- Platform-Targeted WebSearch: 4 to 8 searches across the relevant subset of 9 platforms.
- Result Collection: capture results to a
mkstemptemp file. Apply the WebSearch Untrusted-Data Contract (sanitize snippets for instruction-shaped patterns; never follow directives in fetched content). - Brief Generation:
scripts/discourse_research.pyclusters by theme, classifies into NEW / CONSENSUS / NICHE / SPECIFICS buckets, applies the 6 synthesis-contract LAWs, writesDISCOURSE.mdatomically.
Security note: the script enforces strict input validation (JSON schema,
URL scheme allowlist, control-char stripping, MAX_STRING_FIELD cap) and
TOCTOU-resistant file reads (O_NOFOLLOW on POSIX). See
tests/test_security_v1_8_0.py and the v1.8.1 hardening pass for the
threat model.
/blog write
Write a new blog post from scratch, fully optimized for Google rankings and AI citation platforms.
Usage
/blog write <topic>
/blog write "How to Optimize for AI Search in 2026"
/blog write kubernetes monitoring
/blog write takes only the topic as an argument. Things like the output
format (MDX, Markdown, HTML), a target word count or the audience are not
flags: say them in plain language ("write it as MDX, about 3,000 words, for
platform engineers") and the skill asks for whatever it still needs.
Workflow
- Topic clarification: Asks for audience, keyword, word count, platform
- Research: Spawns
blog-researcheragent to find sourced statistics and images - Outline: Generates structured outline, presents for approval
- Draft: Creates the draft folder and runs Gate 1 capability discovery before writing (with
--reset-iterationsfirst when a new write starts in a used folder), then spawns theblog-writeragent for the full article; the agent may leave[IMAGE],[CHART]and[INTERNAL-LINK]markers - Visuals: The orchestrator resolves every marker (hero image,
blog-chartSVGs where a chart helps, real internal links); none may remain - Render and review: Renders HTML and PDF, issues the review nonce, runs
blog-reviewer, and saves its report toreview.md - Gates: Runs the 5-gate delivery contract; any edit means re-render, new nonce and a new review; after 3 failed gate runs it stops with the diagnostic
- Delivery: Presents the artifact folder and the scorecard
The exact order and every gate check are specified once, in
skills/blog/references/blog-delivery-contract.md.
Output
A complete blog post in the detected format (Markdown, MDX, or HTML) with:
- YAML frontmatter (title, description, coverImage, ogImage, date, tags)
- Purpose-first treatment of important sections, with evidence where needed
- Sourced statistics from tier 1-3 sources (as many as the article needs, no fixed count)
- A local hero image and any inline images the topic needs, from licensed sources
- SVG data visualization charts only where a chart explains the evidence (no fixed count)
- Optional FAQ when genuine reader questions warrant it
- Real internal links (markers must be resolved before rendering)
Related Commands
/blog brief: Generate a brief first, then feed it to/blog write/blog analyze: Score the finished post/blog seo-check: Validate SEO after writing
/blog rewrite
Optimize an existing blog post for rankings and AI citations while preserving the author's voice and unique perspective.
Usage
/blog rewrite <file>
/blog rewrite content/blog/my-post.mdx
/blog rewrite posts/old-article.md
Workflow
- Audit: Reads the file, scores it against the quality checklist
- Plan: Presents section-by-section optimization plan for approval
- Research: Finds replacement statistics for fabricated/unsourced data
- Chart generation: Adds an SVG chart where one would explain the evidence better than prose
- Rewrite: Clarifies important sections, fixes paragraphs, and adds Q&A only when useful
- Verification: Confirms all quality gates pass
- Summary: Reports before/after scores and changes made
Output
The rewritten file in its original format with:
- Truthful
lastUpdatedonly when facts, methods, or recommendations materially changed - Clear, evidence-backed treatment of important sections
- Fabricated statistics replaced with sourced data
- Images and charts added where needed
- FAQ added or improved only when it serves real reader questions
- Self-promotion reduced to max 1 brand mention
Related Commands
/blog analyze: Audit before rewriting to see the starting score/blog update: Alias for freshness-focused rewrite
/blog analyze
Audit a blog post's quality with a 0-100 score across 5 categories, with prioritized improvement recommendations.
Usage
/blog analyze <file>
/blog analyze <url>
/blog analyze content/blog/ --batch
Input Types
| Input | Behavior |
|---|---|
Local file (.md, .mdx, .html) |
Reads and analyzes directly |
| URL | Fetches via WebFetch, extracts content |
Directory (with --batch) |
Scans all blog files, produces summary table |
Output
Blog Quality Report: [Title]
Score: 75/100 - Acceptable
Score Breakdown
| Category | Score | Max |
|------------------------|-------|-----|
| Content Quality | 24 | 30 |
| SEO Optimization | 19 | 25 |
| E-E-A-T Signals | 11 | 15 |
| Technical Elements | 11 | 15 |
| AI Citation Readiness | 10 | 15 |
Issues Found (prioritized: Critical > High > Medium > Low)
Recommended Actions (top 3 highest-impact fixes)
Batch Mode Output
When given a directory, produces a ranked summary table of all posts with scores, ratings, and top issues. Posts are sorted lowest-score-first as a priority queue for optimization.
Python Script
The analyze_blog.py script provides automated metrics:
python3 ~/.claude/skills/blog/scripts/analyze_blog.py post.md
python3 ~/.claude/skills/blog/scripts/analyze_blog.py posts/ --batch
python3 ~/.claude/skills/blog/scripts/analyze_blog.py post.md -o report.json
Related Commands
/blog rewrite: Apply the recommended fixes automatically/blog audit: Full-site assessment (broader than single-file analyze)
/blog brief
Generate a comprehensive content brief with keywords, competitive analysis, statistics research, visual element planning, and a structured outline.
Usage
/blog brief <topic>
/blog brief "cloud cost optimization for startups"
/blog brief ai-search-optimization
There are no audience or other flags. The skill asks for the audience and business context in its first step, or you can mention them in plain language ("brief for marketing managers").
Workflow
- Topic intake: Gathers topic, audience, intent, business context
- Keyword research: Primary keyword, 3-5 secondary, 3-5 questions
- Competitive analysis: Analyzes top 3-5 ranking pages
- Statistics research: Finds 8-12 stats with sources
- Brief generation: Complete brief with outline and recommendations
Output
A detailed brief document saved to briefs/[slug]-brief.md containing:
- Target keywords (primary, secondary, questions)
- Search intent analysis
- Content parameters (word count, format, and image or chart guidance as the topic needs)
- Recommended title options and meta description
- Full content outline with section-level guidance
- Statistics table with sources pre-researched
- Visual element plan (chart types, image search terms)
- Competitive gaps to exploit
- Internal linking opportunities
- E-E-A-T signals to include
- Distribution notes (Reddit, YouTube, social)
Related Commands
/blog write: Write the article using the generated brief/blog strategy: Higher-level planning before individual briefs/blog outline: Lighter-weight outline without full research
/blog calendar
Generate an editorial calendar with topic clusters, publishing schedules, freshness update plans, and seasonal opportunities.
Usage
/blog calendar
/blog calendar monthly
/blog calendar quarterly
/blog calendar "devops tooling"
The optional argument is a niche, as plain text. Publishing cadence is not a flag: say it in plain language ("three posts a week") when the skill asks.
Output Formats
Monthly calendar: Week-by-week table with post type (New/Update), title, topic cluster, target keyword, and status. Includes freshness update queue and seasonal hooks.
Quarterly calendar: Three-month plan with cluster focus per month, content velocity targets, quarterly goals, and distribution plan.
Key Features
- Topic cluster design: 3-5 pillar + supporting content clusters
- Freshness scheduling: 30-day update cycles for high-priority posts
- Content mix: Balances new posts with freshness updates
- Seasonal hooks: Industry events, trending topics, report releases
Related Commands
/blog strategy: Define pillars and positioning before calendar planning/blog brief: Create briefs for calendar items
/blog strategy
Develop a comprehensive blog strategy with content pillars, audience mapping, competitive landscape analysis, and distribution planning.
Usage
/blog strategy <niche>
/blog strategy "B2B SaaS marketing"
/blog strategy ecommerce
Competitors are not a flag: name them in plain language ("compare against Shopify, BigCommerce and WooCommerce") or let the skill ask during discovery.
Workflow
- Discovery: Business context, goals, current state, competitors
- Competitive landscape: Analyzes competitor blogs and AI visibility
- Audience mapping: 2-3 segments with pain points and search behavior
- Content pillar design: 3-5 pillars with keyword themes
- Differentiation: First-hand experience and original data plans
- Distribution channels: YouTube, Reddit, reviews, publications
- Measurement framework: Traditional SEO + AI citation metrics
- Strategy document: Executive summary through 90-day roadmap
Output
A full strategy document with:
- Audience segments with AI behavior analysis
- Content pillars with estimated post counts
- Competitive positioning and gaps
- Distribution channel priorities with tactics
- Content velocity recommendations
- 90-day implementation roadmap
- KPIs across SEO, AI citation, quality, and business impact
Related Commands
/blog calendar: Operationalize the strategy into a publishing schedule/blog brief: Create briefs for strategy-identified topics
/blog outline
Generate a SERP-informed content outline by analyzing what currently ranks for the target keyword.
Usage
/blog outline <topic>
/blog outline "react server components best practices"
Output
A structured outline with:
- H2 section headings in the form that best matches reader intent
- Purpose-first guidance for important sections
- Image and chart placement markers
- FAQ question suggestions
- Word count targets per section
Related Commands
/blog brief: Full brief with research (outline is a subset)/blog write: Write from the outline directly
/blog seo-check
Post-writing SEO validation that checks technical SEO elements beyond content quality.
Usage
/blog seo-check <file>
/blog seo-check content/blog/new-post.mdx
Checks Performed
- Meta title clarity and fit for the page
- Accurate, page-specific meta description
- Heading hierarchy (H1 > H2 > H3, no skips)
- Keyword presence in title, H2s, and meta description
- Internal link count (target 5-10 per 2,000 words)
- Image alt text completeness
- Schema markup presence (BlogPosting, FAQPage)
- Open Graph / Twitter Card meta tags
lastUpdated/dateModifiedtruthfulness after substantive changes
Related Commands
/blog analyze: Full quality audit (content + SEO + citations)/blog schema: Generate missing schema markup
/blog schema
Generate JSON-LD structured data markup for a blog post.
Usage
/blog schema <file>
/blog schema content/blog/my-post.mdx
Generated Schema Types
| Schema Type | When Generated |
|---|---|
| BlogPosting | Always (primary) |
| FAQPage | When FAQ section detected |
| BreadcrumbList | When site structure available |
| Person | When author information available |
| Organization | When company context available |
Output
JSON-LD <script> blocks ready for injection into the page <head> or
component. Includes datePublished, dateModified, author, image,
and FAQ items.
Important
Prefer schema in source or server-rendered HTML for portability because crawler rendering capabilities vary. For Google Search, JavaScript-generated JSON-LD is acceptable when it reaches the rendered DOM, matches visible content, and passes validation. Verify other target crawlers individually.
Related Commands
/blog seo-check: Validates schema presence and correctness/blog analyze: Checks schema as part of the full quality audit
/blog repurpose
Repurpose a blog post into content for other platforms and formats.
Usage
/blog repurpose <file>
/blog repurpose content/blog/ai-search-guide.mdx
Output Formats
| Platform | Format |
|---|---|
| Twitter/X | Thread (5-10 tweets with key insights) |
| Article or post with professional angle | |
| Discussion-starter post for relevant subreddits | |
| YouTube | Video script outline with talking points |
| Newsletter | Email digest version with CTA |
| Podcast | Interview/discussion script based on post content |
Related Commands
/blog strategy: Identifies distribution channels for repurposing/blog write: Create the original post to repurpose
/blog geo
AI citation optimization audit. Analyzes a blog post specifically for visibility on AI platforms (ChatGPT, Perplexity, Google AI Overviews).
Usage
/blog geo <file>
/blog geo content/blog/my-post.mdx
Checks Performed
- Evidence-backed, self-contained treatment of important claims
- Substantive maintenance when facts, methods, or recommendations change
- Optional visible Q&A with no Google rich-result or readiness-score benefit
- Source authority tier quality
- Reader-useful structure and enough context for important claims to stand alone
- JavaScript dependency and target-crawler rendering compatibility
robots.txtAI crawler access (GPTBot, ClaudeBot, PerplexityBot)- Optional audience-channel recommendations (YouTube, Reddit, reviews)
Related Commands
/blog analyze: Full quality audit including GEO metrics/blog rewrite: Apply GEO optimizations automatically
/blog audit
Full-site blog health assessment. Scans an entire directory of blog posts and produces a comprehensive report.
Usage
/blog audit
/blog audit content/blog/
/blog audit posts/ --format markdown
Output
- Summary table: Every post scored and rated
- Priority queue: Posts ranked lowest-score-first for optimization
- Category breakdown: Site-wide averages per scoring category
- Common issues: Most frequent problems across all posts
- Freshness report: Posts overdue for updates (>30 days)
- Topic coverage: Cluster analysis and gap identification
Related Commands
/blog analyze: Single-file audit (audit is the batch version)/blog calendar: Plan content based on audit findings/blog rewrite: Fix posts identified by the audit
/blog update
Alias for /blog rewrite with a freshness-focused mode. Minimizes structural
changes and focuses on updating data and signals.
Usage
/blog update <file>
/blog update content/blog/old-post.mdx
What It Does
- Updates statistics to latest available data (2025-2026)
- Adds new developments since last update
- Refreshes images if older than 1 year
- Updates
lastUpdatedonly after substantive content changes - Preserves existing structure (minimal rewrites)
- Makes only the changes needed for factual accuracy and reader value; no percentage-change target
Related Commands
/blog rewrite: Full rewrite (more aggressive than update)/blog audit: Find posts that need updating
/blog image
Purpose: AI image generation and editing via the nanobanana-mcp server (Gemini).
/blog image generate <description>
/blog image edit <path> <instructions>
/blog image setup
Flow: See skills/blog-image/SKILL.md: Creative Director pattern with
the 6-component Reasoning Brief (Subject, Action, Context, Composition,
Lighting, Style). The setup subcommand defaults to --global (writes
user-private ~/.claude/settings.json, mode 0600) per audit VULN-001 fix.
/blog cannibalization
Purpose: detect keyword overlap across posts in a directory.
/blog cannibalization [directory]
Flow: See skills/blog-cannibalization/SKILL.md.
/blog factcheck
Purpose: verify statistics in a draft against cited sources.
/blog factcheck <file-path>
Flow: See skills/blog-factcheck/SKILL.md.
/blog persona
Purpose: manage writing personas and voice profiles.
/blog persona create
/blog persona list
/blog persona use <name>
/blog persona show <name>
Flow: See skills/blog-persona/SKILL.md. Personas are saved in your
project at <project-root>/.claude-blog/personas/<name>.json, next to
BRAND.md and VOICE.md, never inside the installed plugin (a plugin update
replaces that directory). Files already in the legacy
skills/blog-persona/references/personas/ directory are still listed, and you
can copy them to the project directory.
/blog taxonomy
Purpose: tag/category CMS management for blog content.
/blog taxonomy sync
/blog taxonomy audit
/blog taxonomy suggest
Flow: See skills/blog-taxonomy/SKILL.md.
/blog notebooklm
Purpose: query NotebookLM (renamed Gemini Notebook; notebook URLs are on
notebook.google.com, and the legacy notebooklm.google.com host is still
accepted) for source-grounded research via patchright browser automation.
/blog notebooklm ask <question>
/blog notebooklm discover <url>
/blog notebooklm library [list|add <url>|search <query>|remove <id>]
/blog notebooklm setup
/blog notebooklm status
/blog notebooklm cleanup
Flow: See skills/blog-notebooklm/SKILL.md. With --json, the notebook's
answer is returned as untrusted data and the follow-up prompt arrives
separately in follow_up_reminder. Cookies at
~/.claude/skills/blog-notebooklm/data/ written mode 0600 per audit
VULN-004 fix.
/blog audio
Purpose: generate audio narration via Gemini TTS.
/blog audio generate <file-path>
/blog audio voices
/blog audio setup
Flow: See skills/blog-audio/SKILL.md. 30-voice catalog in references.
generate_audio.py defaults to the flash38 model alias, and
--max-chunks (default 12) refuses runs that would make more billed API calls
than you intended. Run generate_audio.py --help for the current model
aliases and which are deprecated.
/blog google
Purpose: Google API data integration (PSI, CrUX, GSC, GA4, NLP, YouTube, Keywords). Indexing API use is limited to JobPosting or livestream URLs.
/blog google <command> [args]
Flow: See skills/blog-google/SKILL.md. 13 commands across 4 credential
tiers. OAuth state CSRF + chmod 0600 token storage per audit VULN-002 +
VULN-008 fixes. Default scopes are read-only (gsc_readonly + ga4); pass
--scopes to elevate. gsc_query.py --days N sets the window length (default
28 days, ending 3 days ago because of Search Console data lag).
indexing_notify.py refuses to submit anything unless you pass
--confirm-eligible, which asserts that every URL is a JobPosting page or a
BroadcastEvent-in-VideoObject page, the only types the Indexing API supports;
ordinary blog posts are not eligible.
/blog cluster (v1.7.0)
Purpose: semantic topic-cluster planning + execution. Hub-and-spoke architecture with shared cluster context.
/blog cluster
/blog cluster plan <seed-keyword>
/blog cluster plan --from strategy [path]
/blog cluster execute [path-to-plan]
Flow: See skills/blog-cluster/SKILL.md. Adapted from
semantic-cluster-engine (AI Marketing Hub Pro Hub Challenge winner; the
original repository is no longer available as of 2026-10-12).
/blog flow (v1.7.0)
Purpose: FLOW framework prompts (find / optimize / win / prompts / sync).
/blog flow find
/blog flow optimize
/blog flow win
/blog flow prompts
/blog flow sync [--ref <sha>] [--allow-drift]
Flow: See skills/blog-flow/SKILL.md. Sync uses scripts/sync_flow.py
with HTTPS-only host allowlist + 5MB cap + path-traversal guard + blocking
lockfile drift gate (per audit VULN-018 fix).
/blog multilingual (v1.7.0)
Purpose: one-command write + translate + localize + emit hreflang.
/blog multilingual <topic> --languages de,fr,es
Flow: See skills/blog-multilingual/SKILL.md. Orchestrates blog-write,
blog-translate, blog-localize, plus optional seo-hreflang.
/blog translate (v1.7.0)
Purpose: SEO-optimized translation with format preservation.
/blog translate <file-path> --to <codes>
Flow: See skills/blog-translate/SKILL.md. Spawns blog-translator
agent (least-privilege: Read/Write/Edit/Glob/Grep, no Bash, no Web).
/blog localize (v1.7.0)
Purpose: cultural deep-adaptation per locale (DACH, Francophone, Hispanic, Japanese profiles + custom template).
/blog localize <file-path> --locale <code>
Flow: See skills/blog-localize/SKILL.md.
/blog locale-audit (v1.7.0)
Purpose: multilingual content QA (completeness matrix, hreflang correctness, meta-tag parity, freshness).
/blog locale-audit <directory>
Flow: See skills/blog-locale-audit/SKILL.md.
/blog style
Purpose: learn a measurable author voice profile from 5 to 10 existing
posts, to feed VOICE.md, /blog persona and /blog write. The profile
describes style (sentence length, vocabulary, transitions, phrasing); it does
not detect or attribute authorship.
/blog style learn <paths>
The learner is scripts/style_learn.py; it accepts files or directories and
the options --min, --format {json,markdown} and --output. Fewer samples
than the minimum (default 5) produce a warning, not a failure.
Flow: See skills/blog-style/SKILL.md.
/blog decay
Purpose: compare two Google Search Console page exports (current and previous period) and flag pages whose clicks or impressions fell by at least a threshold (default 20%), with a triage hint. The threshold is a screening heuristic, not proof of decay.
/blog decay <current-gsc.json> <previous-gsc.json>
/blog decay <current-gsc.json> <previous-gsc.json> --threshold 0.30 --metric impressions
The analyzer is scripts/content_decay.py (--threshold, --metric {clicks,impressions}, --format {json,markdown}, --output). Both exports
must come from the same property and use one URL form.
Flow: See skills/blog-decay/SKILL.md.
Command Routing
The main orchestrator (skills/blog/SKILL.md) parses user input and routes to
the correct sub-skill:
User Input Routes To
----------- ---------
/blog write <topic> --> blog-write
/blog rewrite <file> --> blog-rewrite
/blog analyze <file-or-url> --> blog-analyze
/blog brief <topic> --> blog-brief
/blog calendar [monthly|quarterly] [niche] --> blog-calendar
/blog plan [monthly|quarterly] [niche] --> blog-calendar
/blog strategy <niche> --> blog-strategy
/blog ideation <niche> --> blog-strategy
/blog outline <topic> --> blog-outline
/blog seo-check <file> --> blog-seo-check
/blog schema <file> --> blog-schema
/blog repurpose <file> --> blog-repurpose
/blog geo <file> --> blog-geo
/blog audit [directory] --> blog-audit
/blog image [generate|edit] --> blog-image
/blog update <file> --> blog-rewrite (freshness mode)
/blog cannibalization [dir] --> blog-cannibalization
/blog factcheck <file> --> blog-factcheck
/blog persona [create|list|use|show] --> blog-persona
/blog brand [init|show|update] --> blog-brand (v1.8.0)
/blog discourse <topic> --> blog-discourse (v1.8.0)
/blog taxonomy [suggest|sync|audit] --> blog-taxonomy
/blog notebooklm [ask|discover|...] <arg> --> blog-notebooklm
/blog audio [generate|voices|setup] --> blog-audio
/blog google [command] [args] --> blog-google
/blog cluster [plan|execute] [seed-or-plan] --> blog-cluster
/blog multilingual <topic> --languages --> blog-multilingual
/blog translate <file> --to --> blog-translate
/blog localize <file> --locale --> blog-localize
/blog locale-audit <directory> --> blog-locale-audit
/blog style learn <paths> --> blog-style (v1.10.0)
/blog decay <current> <previous> --> blog-decay (v1.10.0)
/blog flow [find|optimize|win|prompts|sync] --> blog-flow
If no sub-command is provided, the orchestrator asks which action the user needs.