publishing-skills

Four composable skills that turn an AI coding agent (Claude, Cursor, Codex, Gemini, Copilot, ...) into a long-tail SEO publishing pipeline. Platform-agnostic - works against Ghost, WordPress, Webflow, Sanity, Strapi, or any static-site generator (Hugo, Astro, Eleventy, Jekyll, Next-MDX).

Built for indie hackers, founders, content marketers, and dev-tool teams who want AI to draft blog posts that actually rank in search and get quoted by AI assistants - not paraphrased docs, not hallucinated benchmarks, not generic "ultimate guide" filler.

| Skill | What it does | Who it's for | |---|---|---| | blog-topic-research | Validates a topic has real, verifiable demand (People Also Ask, Reddit, Stack Overflow, GitHub issues, vendor forums, changelogs) before you spend tokens drafting. Every accepted topic carries citable evidence URLs, a problem summary, confirmed fixes, version context, and FAQ variants the writer can use directly. | Anyone tired of writing posts nobody searches for, and editorial pipelines that need an evidence-backed backlog. | | seo-blog-writer | End-to-end pipeline for a single post: classify → research → outbound interlinks → draft clean HTML → scrub LLM tells → AI-SEO audit → optional glossary auto-link → publish. Adds FAQPage + BreadcrumbList + HowTo JSON-LD for AI-citation extractability. Pre-publish gate asserts H2-question shape, figure count (max(1, words // 500) for 800+ word posts), bullet discipline (3-9 items), and currency (as of <YYYY> qualifier on stale years). Ships a glossary auto-linker that wraps the first mention of each known technical term in an internal link with hover-tooltip metadata. Platform-pluggable publish step - ships with Ghost Admin API, WordPress REST, and static-site adapters; any other CMS is a ~20-line snippet. | Founders and marketers who want to ship one ranking post a day without paying a writer or a designer. | | blog-figure-svg | Generates accessible SVG figures (flow, comparison bars, taxonomy, terminal mocks, 1600x840 OG feature cards) with a consistent palette, screen-reader metadata, and figcaption-ready output. Rasterizes to compressed PNG for upload to any CMS. | Anyone shipping more than 3 posts a month who doesn't want stock photos or Midjourney filler on every article. | | blog-editorial-calendar | The orchestration layer over the other three. Keeps an evidence-backed backlog, picks the next topic so your corpus drifts toward the cluster + format mix you defined in config.json, schedules posts into a rolling daily cadence, reconciles the backlog against what's live on your CMS, and auto-refills via blog-topic-research when the queue runs dry. Drives whatever adapter seo-blog-writer is configured for. | Teams who want a hands-off cadence that publishes on schedule and stays balanced - not a pile of posts in one category. |

Together, they form a complete pipeline: plan the cadence → research the topic → write the post → illustrate it → publish to your platform of choice.

Installing

Via skills.sh (any agent runtime)

Installs all 4 skills into Claude Code, Cursor, Codex, Gemini CLI, GitHub Copilot, and 50+ other runtimes at once:

npx skills add AutomateLab-tech/publishing-skills

Via clawhub CLI (Claude / OpenClaw runtimes)

npm i -g clawhub
clawhub login
clawhub skill install blog-topic-research
clawhub skill install seo-blog-writer
clawhub skill install blog-figure-svg
clawhub skill install blog-editorial-calendar

Manually

Each skill is a single SKILL.md file with YAML frontmatter. Drop it into your project's .claude/skills/<name>/ directory (or your agent runtime's equivalent skill folder) and reload.

git clone https://github.com/AutomateLab-tech/publishing-skills.git
cp -r publishing-skills/skills/* .claude/skills/

How they compose

+----------------------+      +--------------------------+      +-------------------+      +-------------------+
| blog-topic-research  | ---> | blog-editorial-calendar  | ---> |  seo-blog-writer  | <--- |  blog-figure-svg  |
| 'is the topic worth  |      | 'pick the next topic to  |      | 'write + publish' |      | 'illustrate the   |
|  writing about?'     | <--- |  balance the corpus +    |      |                   |      |   post'           |
+----------------------+ refill  schedule the cadence'   |      +-------------------+      +-------------------+
                              +--------------------------+               |
                                                                         v
                                              +----------------------------------+
                                              | platform adapter (one of):       |
                                              |   - Ghost Admin API              |
                                              |   - WordPress REST API           |
                                              |   - static-site file output      |
                                              |   - bring-your-own (~20 lines)   |
                                              +----------------------------------+

• blog-topic-research runs first - it answers "does anyone actually search this?" before you spend tokens drafting.
• blog-editorial-calendar picks the next topic to keep the corpus balanced against your target mix, schedules it into a rolling cadence, and calls blog-topic-research to refill when the backlog runs dry. (Optional - you can drive the writer by hand instead.)
• seo-blog-writer consumes the research-proof scaffold (or runs from scratch), drafts the HTML, validates the bundle, and ships it through a platform adapter.
• blog-figure-svg is invoked from within the writer's illustration step (or standalone) when a post needs charts, diagrams, or a feature card.

Requirements

• blog-topic-research - no system dependencies beyond WebSearch / WebFetch in the agent runtime.
• seo-blog-writer - Python 3. Platform-specific extras:
  • Ghost adapter: pip install requests pyjwt; GHOST_URL + GHOST_ADMIN_KEY env vars.
  • WordPress adapter: pip install requests; WP_URL + WP_USER + WP_APP_PASSWORD env vars.
  • Static-site adapter: no credentials; just a target directory in your SSG repo.
• blog-figure-svg - Python 3, plus one SVG rasterizer (ImageMagick / rsvg-convert / Inkscape / cairosvg) and optionally pngquant for compression.

Optional - ai-seo MCP (programmatic citation scoring)

seo-blog-writer Step 5 (AI-SEO audit) runs a programmatic citation-worthiness and schema pass when the ai-seo MCP (@automatelab/ai-seo-mcp) is connected. The skill checks at preflight and prompts you to install it if it's missing. Without it, Step 5 falls back to a manual reasoning pass covering the same ground.

npx -y @automatelab/ai-seo-mcp

Then register it in your agent's MCP config. See the ai-seo-mcp README for one-line configs for Claude Code, Cursor, and Cline.

Maintenance scripts

The per-post audit inside seo-blog-writer catches structural problems before publish. For corpus-wide drift - characters or banlist phrases that crept back across many posts - run scripts/audit-corpus.py against your content directory:

python3 scripts/audit-corpus.py path/to/your/content/
python3 scripts/audit-corpus.py content/posts/ --extra "synergy,best-in-class"

Exits 0 clean / 1 on hits - composes with CI for a pre-deploy drift gate.

Glossary auto-link

seo-blog-writer Step 7i (optional) pipes the draft HTML through scripts/inject-glossary-links.py, which wraps the first mention of each known technical term in an internal link to its definition page and writes a data-definition attribute for hover tooltips. Schema and starter file in skills/seo-blog-writer/references/glossary-schema.md; a self-contained tooltip-rendering snippet for your site <head> is at skills/seo-blog-writer/references/decorate.js.

python3 scripts/inject-glossary-links.py post.html \
    --glossary glossary.json --base-url /glossary/ > post.linked.html

The injector handles the awkward edges automatically: first-occurrence-only per post, longest-alias wins, skips headings / code / tables / blockquotes / TL;DR, and rejects matches embedded in identifier-like compounds (user-agent won't match agent, @scope/ai-seo-mcp won't match mcp). You bring the glossary file and the definition pages; the injector handles the linking.

Why platform-agnostic?

The writing pipeline (research, classify, draft, scrub, AI-SEO audit, JSON-LD generation) is the hard part. The publish step is glue. Coupling the two means you're locked into one CMS forever - and if you move from Ghost to WordPress, you rewrite the whole skill.

This repo separates them. The writer produces a stable bundle (<slug>.draft.html, <slug>.schema.html, <slug>.metadata.json) and the adapter ships it. Add Webflow, Sanity, or Strapi in an afternoon without touching the pipeline.

License

MIT-0 - public domain equivalent. Use, modify, redistribute without attribution.