Blog Generator

What this skill does

• Scans journal entries from the last N days for interesting topics (discoveries, obstacles, solutions)
• Identifies high-value topics based on keyword relevance and problem-solving value
• Researches search volume and keyword opportunities (heuristic-based, can be enhanced with APIs)
• Generates structured blog posts with overview, problem, solution, and takeaways sections
• Saves blog posts to /Users/ghost/.openclaw/blogs/ as HTML only (X-article format, header from visual-explainer, humanizer between generations)

Integration as a Cron Job

This skill is designed to run periodically (daily or weekly) via OpenClaw cron to automatically generate blog content.

Example Cron Job Configuration (Daily):

{
  "payload": {
    "kind": "agentTurn",
    "message": "Run blog-generator skill to analyze journal entries and generate high-value blog posts.",
    "model": "openrouter/google/gemini-2.5-flash",
    "thinking": "low",
    "timeoutSeconds": 300
  },
  "schedule": {
    "kind": "cron",
    "cron": "0 9 * * *"
  },
  "delivery": {
    "mode": "announce"
  },
  "sessionTarget": "isolated",
  "name": "Blog Post Generator"
}

Example Cron Job Configuration (Weekly):

{
  "payload": {
    "kind": "agentTurn",
    "message": "Run blog-generator skill with --days 7 --max-topics 3 to generate weekly blog posts from journal analysis.",
    "model": "openrouter/google/gemini-2.5-flash",
    "thinking": "low",
    "timeoutSeconds": 300
  },
  "schedule": {
    "kind": "cron",
    "cron": "0 10 * * 1"
  },
  "delivery": {
    "mode": "announce"
  },
  "sessionTarget": "isolated",
  "name": "Weekly Blog Generator"
}

Output Format

HTML only (no Markdown). Output is in the format accepted by X articles: one header (from visual-explainer), then article body with no interlaced visuals.

• File: /Users/ghost/.openclaw/blogs/YYYYMMDD_slugified-title.html
• Article structure: X-style (long-form): punchy hook, short paragraphs, "two types of people" framing, pivot, stakes. Use --format classic for overview/problem/solution.
• Header: From visual-explainer only. Default path: /Users/ghost/.openclaw/workspace/skills/visual-explainer-main. The skill’s scripts/generate_header.py is called with section: "header" and returns an html_snippet (hero with title and optional summary). No diagrams or images in the body.
• Humanizer: Runs between generations by default. Each article’s body is sent through /Users/ghost/Downloads/humanizer-1.0.0 (requires OPENROUTER_API_KEY) before rendering to HTML. Use --no-humanize to skip.

Topic Scoring

Topics are scored based on:

• High-value keywords: OpenClaw-specific terms, problem-solving language
• Content type: Solutions score highest, then obstacles, then discoveries
• Content depth: Longer, more detailed content scores higher
• Search volume indicators: Keywords like "how to", "tutorial", "fix" increase value

Requirements

• Journal entries in /Users/ghost/.openclaw/journal/
• Blogs directory writable at /Users/ghost/.openclaw/blogs/
• Chat history analyzer skill (for journal entries)

How it works

1. Scans journal directory for markdown files from the last N days
2. Extracts topics from discoveries, obstacles, and solutions sections
3. Scores topics based on keyword relevance and value
4. Selects top N high-value topics
5. Generates structured blog posts with problem/solution format
6. Saves posts to blogs directory with timestamped filenames

Enhancement Opportunities

• Integrate with Google Keyword Planner API for real search volume data
• Use AI model to enhance blog post quality and SEO optimization
• Cross-reference with existing blog posts to avoid duplicates
• Generate multiple variations of posts for A/B testing