cm-auto-publisher: The Agentic Publishing Skill

SkillDocs & knowledge

Comprehensive publishing automation skill bridging AI agents (OpenClaw, Manus, Claude) to the Content Factory Router. Uses the Router API to publish markdown articles with rich media to any connected Astro site natively.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the cm-auto-publisher: The Agentic Publishing Skill skill

What this skill tells your AI

The instructions your AI receives, as published by tody-agent/codymaster in skills/cm-auto-publisher/SKILL.md and read by ahel’s review.

Goal: Take unstructured input (links, ideas, raw text), write a highly engaging article following Content Factory rules, and instantly publish it via the Content Factory Router API.

When to Use

ALWAYS when:

  • User provides a link/video and says "Write an article about this and publish it".
  • User says "Draft and push this news to the blog".
  • You have finished drafting an article and need to deploy it to the live site.
  • An automated cron job triggers you to aggregate news and auto-publish.

The Process

Phase 1: Content Extraction & Structuring

  1. Analyze Input: If given a URL, use read_url_content (or your browsing tool) to read the source. If given images or Youtube links, extract them.
  2. Draft Content: Write a compelling article in standard Markdown.
    • Use headings (##, ###).
    • For YouTube videos, use the Astro component: <YouTube id="VIDEO_ID" />.
    • For images, just use standard markdown: ![Alt text](image_filename.jpg).

Phase 2: Media Preparation

Identify all external images that need to be published with the article. DO NOT try to download them to the local disk using bash commands. The Content Factory Router will handle this. Instead, build a media array mapping the original URLs to the desired clean filenames.

"media": [
  {
    "url": "https://source.com/raw-image-123.jpg",
    "filename": "clean-seo-friendly-name.jpg"
  }
]

Phase 3: API Payload Construction

Construct the JSON payload matching the target site's schema. Standard Target: cody-master (mapped to tody-agent/todyle-web).

{
  "site_id": "cody-master",
  "title": "[Engaging Title]",
  "description": "[1-2 sentence SEO meta description]",
  "category": "news", // or "guide", "use-case"
  "readingTime": "5 min read",
  "tags": ["AI", "Vibe Coding", "Automation"],
  "content": "The raw markdown content here...",
  "media": [ ... ]
}

Phase 4: Publish via Router API

Send the payload to the Cloudflare Worker via a standard fetch or curl command.

  1. Get the Secret: Retrieve the API token from the .env file or ask the user for the PUBLISHER_API_KEY.
  2. Send POST Request: Use a local script or curl to send the data to the worker.
# Example curl (assuming the payload is saved to payload.json)
curl -X POST https://content-factory-router.<YOUR-CLOUDFLARE-ACCOUNT>.workers.dev/publish \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <API_KEY>" \
  -d @payload.json

Phase 5: Verification

  1. Check the response from the Worker. It should return {"success": true, "message": "Published...", "commit_sha": "..."}.
  2. Inform the user that the article has been successfully committed to GitHub and Cloudflare Pages is now building it.

Red Flags - STOP!

  • DO NOT manually create .mdx files in src/content/articles/ and run git push if using this skill! Let the Router API handle the Git history and image blob creation to ensure data consistency.
  • DO NOT use cat or echo to build complex JSON payloads in bash. Save the payload to a .json file using your file writing tool, then pass it to curl.

Signals

GitHub stars
52
Forks
23
Last commit
Aug 2026
Advanced
Catalog kind
skill
Gateway key
cm-auto-publisher
Source
github.com/tody-agent/codymaster