knowledge-base-publisher

Knowledge Base Publisher

A skill for turning raw content into structured, publication-ready knowledge base articles.

When to Use

• Converting meeting notes into structured documentation
• Organizing scattered content into a knowledge base
• Adding metadata, tags, and cross-references to documentation
• Batch-processing multiple articles for publication
• Versioning knowledge base entries with changelog tracking

Prerequisites

• A source directory containing raw content (markdown, text, or transcript files)
• A target directory or platform for published output
• Optional: a taxonomy file for consistent tagging

Basic Steps

1. Ingest Raw Content

Read source files from the input directory. Supports:

• .md / .markdown files
• .txt plain text files
• Transcript formats (JSON with speaker/timestamp)

2. Structure & Enrich

For each document:

1. Extract or generate title — from H1, filename, or AI summary
2. Add metadata block — date, author, tags, related articles
3. Normalize formatting — consistent headings, code blocks, lists
4. Add cross-references — link to related articles by tag/topic matching
5. Generate summary — one-paragraph abstract if none exists

3. Quality Gate

Check each article for:

• Minimum length (recommend ≥ 200 words)
• Has a clear title and at least one section
• No broken internal links
• Tags match the taxonomy (if provided)

4. Publish

Write structured articles to the output directory with:

• Consistent filename slugification
• index.json catalog with all article metadata
• CHANGELOG.md tracking version history
• Per-article meta.json for machine-readable metadata

Example Output Structure

knowledge-base/
├── index.json              # Full catalog of all articles
├── CHANGELOG.md            # Version history
├── articles/
│   ├── getting-started.md
│   ├── getting-started.meta.json
│   ├── api-reference.md
│   ├── api-reference.meta.json
│   └── ...
└── taxonomy.json           # Tag/category definitions

Taxonomy Format (Optional)

{
  "categories": ["tutorial", "reference", "guide", "faq"],
  "tags": ["api", "setup", "configuration", "troubleshooting"],
  "related": {
    "api": ["getting-started", "authentication"],
    "setup": ["configuration"]
  }
}

Integration with TaskFlow

This skill works well as a TaskFlow child task:

1. Parent flow defines the source directory and taxonomy
2. Spawn this skill as a detached child task for batch processing
3. Flow waits on child completion, then validates the output catalog
4. Flow finishes with a summary of articles published

Integration with ClawHub

Published knowledge bases can be packaged as ClawHub skills:

1. Structure the knowledge base as a skill folder with SKILL.md
2. Use clawhub publish to register it in the ClawHub registry
3. Others can discover and install via clawhub search + clawhub install