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.
Python agent-skillsaiai-citationsai-contentai-marketingai-marketing-hubblogblog-writingclaude-codeclaude-code-skillclaude-pluginclaude-skillcontent-creationcontent-optimizationcontent-strategyeeatgeomultilingualopen-sourceseo
docs: correct counts, validation steps and command sections
Agrici Daniel320665fe2da4
diff --git a/.github/CONTRIBUTING.md b/.github/CONTRIBUTING.mdindex 6268684..7f48d5c 100644--- a/.github/CONTRIBUTING.md+++ b/.github/CONTRIBUTING.md@@ -7,7 +7,9 @@ Thank you for your interest in contributing to claude-blog!1. Fork and clone the repository2. Install dev dependencies: `pip install -e ".[dev]"`3. Run tests: `python -m pytest tests/ -v`-4. Validate plugin metadata: `claude plugin validate .`+4. Validate plugin metadata: `python3 .github/scripts/validate_plugin_tolerate_reserved_name.py`+ (plain `claude plugin validate .` on Claude Code 2.1.292 or later reports the+ reserved `claude-` name once per manifest; the wrapper tolerates exactly that)## Development@@ -38,7 +40,7 @@ claude-blog/- **Template paths** in sub-skills use `templates/` (relative to installed location).- **Agent frontmatter** must NOT include `Bash` in the `tools` list (blast-radius reduction for prompt-injection surfaces; enforced by tests).- Run `python -m pytest tests/ -v` before submitting.-- Run `claude plugin validate .` before submitting.+- Run the plugin validation wrapper from step 4 before submitting.### Code Style@@ -77,7 +79,7 @@ claude-blog/1. Create a feature branch from `main`.2. Make your changes with clear conventional-commit messages.3. Ensure all tests pass (`python -m pytest tests/`).-4. Ensure plugin validates (`claude plugin validate .`).+4. Ensure plugin validates (the wrapper from the Getting Started steps).5. Update `CHANGELOG.md` with an `[Unreleased]` entry describing the change.6. Submit a PR using the template. Fill in: type of change, linked issue, test plan, docs checklist, and security checklist (if applicable).diff --git a/.gitignore b/.gitignoreindex 2a5e210..f5dfa31 100644--- a/.gitignore+++ b/.gitignore@@ -61,7 +61,7 @@ firebase-debug.log# Generated outputoutput/demo-output/-.raw/+/.raw/# Working documentsdefinitive-resesarch-blueprint-for-claude-blog.mddiff --git a/CLAUDE.md b/CLAUDE.mdindex b802052..c15dcc7 100644--- a/CLAUDE.md+++ b/CLAUDE.md@@ -67,17 +67,21 @@ claude-blog/blog-strategy/SKILL.md # Blog positioning and planningblog-seo-check/SKILL.md # Post-writing SEO validationblog-schema/SKILL.md # JSON-LD schema generation- blog-chart/SKILL.md # Inline SVG data visualizations+ blog-chart/ # Inline SVG data visualizations+ SKILL.md+ references/ # 1 reference doc (chart input schema)blog-repurpose/SKILL.md # Multi-platform repurposingblog-geo/SKILL.md # AI citation optimizationblog-audit/SKILL.md # Full-site blog health assessmentblog-image/ # AI image generation via GeminiSKILL.md # Image generation sub-skill- references/ # 3 reference docs (models, tools, prompts)+ references/ # 4 reference docs (models, tools, prompts, MCP route)scripts/ # MCP setup and validation scriptsblog-cannibalization/SKILL.md # Keyword overlap detectionblog-factcheck/SKILL.md # Statistics verification- blog-persona/SKILL.md # Writing persona management+ blog-persona/ # Writing persona management+ SKILL.md+ references/ # personas/ directory for saved personasblog-taxonomy/SKILL.md # CMS taxonomy managementblog-notebooklm/ # NotebookLM source-grounded researchSKILL.md # NotebookLM query sub-skill@@ -85,7 +89,7 @@ claude-blog/scripts/ # 11 Python modules + requirements.txt/lockblog-audio/ # Audio narration via Gemini TTSSKILL.md # Audio generation sub-skill- references/ # 1 reference doc (30 voice catalog)+ references/ # 2 reference docs (30 voice catalog, dated models and pricing)scripts/ # 5 Python scripts + requirements.txtblog-google/ # Google API integrationSKILL.md # Google API sub-skill (13 commands, 4 tiers)@@ -98,6 +102,7 @@ claude-blog/blog-flow/ # FLOW framework prompts (v1.7.0)SKILL.md # FLOW orchestrator (find/optimize/win/prompts/sync)references/ # Synced from github.com/AgriciDaniel/flow (CC BY 4.0)+ scripts/ # sync_flow.py wrapper around the core scripts/sync_flow.pyblog-multilingual/ # One-command international publishing (v1.7.0)SKILL.md # Multilingual orchestratorblog-translate/ # SEO-optimized translation (v1.7.0)diff --git a/docs/ARCHITECTURE.md b/docs/ARCHITECTURE.mdindex a7f289b..5a4e916 100644--- a/docs/ARCHITECTURE.md+++ b/docs/ARCHITECTURE.md@@ -216,7 +216,7 @@ dependencies are stdlib-only or narrowly pinned.| `dependency_smoke.py` | Offline initialization smoke checks for the optional runtimes | Maintenance || `discourse_research.py` | Discourse-brief synthesis from SERP JSON; depth-bounded parsing; path-traversal guards | v1.8.0 || `generate_hero.py` | Hero image ladder: Banana MCP -> Gemini API -> Unsplash/Pexels/Pixabay -> Openverse | v1.9.0 |-| `installer_ownership.py` | Shared Unix legacy/v2 ownership verification, migration and rollback | Unreleased |+| `installer_ownership.py` | Shared Unix legacy/v2 ownership verification, migration and rollback | v2.3.0 || `lint_prose.py` | Fence-aware prose-hygiene linter (no em-dash, en-dash, ` -- `); CI-enforced | v1.8.4 || `load_untrusted_root.py` | Code-enforced BRAND/VOICE/DISCOURSE fencing with CSPRNG nonces; O_NOFOLLOW + size cap | v1.8.3 || `quality_gate.py` | Pre-commit gate: block posts scoring < 70; fail-closed on bad paths and missing frontmatter; `--select` is the pre-commit hook mode that skips Markdown files that are not blog posts | v1.10.0 |@@ -278,8 +278,8 @@ dependencies are stdlib-only or narrowly pinned.+-- Writes file to user's project|v- Delivery summary: md, html, pdf, hero, 3 viewport screenshots (preview/),- review.md, preflight-report.json+ Delivery summary: md, html, pdf, hero, 3 viewport screenshots plus a dark-mode+ screenshot (preview/), review.md, preflight-report.json```The authoritative sequence and the full list of what each gate enforces livediff --git a/docs/COMMANDS.md b/docs/COMMANDS.mdindex 547a0b8..a0e2872 100644--- a/docs/COMMANDS.md+++ b/docs/COMMANDS.md@@ -6,11 +6,9 @@ blog-write/blog-rewrite.Every command is invoked through the main orchestrator(`skills/blog/SKILL.md`), which routes to the appropriate sub-skill.-> **For detailed command flows beyond the overview table below, see each-> sub-skill's `SKILL.md` directly. Sections in this file are abbreviated-> for the v1.7.0 and v1.8.0 commands that joined after this doc was-> originally written; the full overview table below is the canonical-> command list.**+> **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@@ -854,6 +852,45 @@ correctness, meta-tag parity, freshness).---+## /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 RoutingThe main orchestrator (`skills/blog/SKILL.md`) parses user input and routes to@@ -892,6 +929,8 @@ User Input Routes To/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```diff --git a/docs/DEMO.md b/docs/DEMO.mdindex a43f000..8d2cb64 100644--- a/docs/DEMO.md+++ b/docs/DEMO.md@@ -135,7 +135,7 @@ here.Output: file path under `~/Documents/nanobanana_generated/`. Bananaacts as Creative Director and constructs the 5-component prompt;the skill auto-loads `references/gemini-models.md` and-`references/prompt-engineering.md` per its MANDATORY rule.+`references/prompt-engineering-blog.md` per its MANDATORY rule.### B2. "Inline SVG chart from data"diff --git a/docs/INSTALLATION.md b/docs/INSTALLATION.mdindex 93fdff9..0051d72 100644--- a/docs/INSTALLATION.md+++ b/docs/INSTALLATION.md@@ -160,12 +160,8 @@ If you prefer not to run the installer, copy files to these paths manually.├── skills/│ ├── blog/│ │ ├── SKILL.md # Main orchestrator-│ │ ├── references/-│ │ │ ├── content-rules.md-│ │ │ ├── geo-optimization.md-│ │ │ ├── google-landscape-2026.md-│ │ │ ├── quality-scoring.md-│ │ │ └── visual-media.md+│ │ ├── references/ # 22 files; copy the whole directory+│ │ │ └── *.md│ │ ├── templates/ # 12 content type templates│ │ │ └── *.md│ │ └── scripts/