# Learning Library - AI Agent Guide
# Last updated: 2026-01-09
# Total items: 1832

## Description
A curated learning library containing educational content on technology, programming, AI/ML, security, and more.

Content types:
- Videos: 1719 (YouTube transcripts with timestamps)
- Papers: 25 (Research papers with abstracts)
- Podcasts: 2 (Audio transcripts)
- Blogs: 2 (Technical articles)

Topics: ai-ml, ai-safety, career, computer-vision, databases, devops, efficiency, entrepreneurship, legal, multimodal, nlp, other, programming, reinforcement-learning, robotics, security, web-development

## Quick Access

Data:
- /library.json - Complete index with all metadata (1832 items)
- /sitemap.xml - All page URLs
- /.well-known/ai.json - Machine-readable API capabilities

## REST API

Base URL: https://youtube-library-docent.dlkarpay.workers.dev

Endpoints:
- GET /api/search?q=<query>&type=video|paper|podcast|blog|all&topic=<topic>&limit=20
- GET /api/recommend?topic=<topic>&level=beginner|intermediate|advanced&limit=10
- GET /api/learning-path?goal=<learning_goal>
- GET /api/whats-new?days=7&type=all
- GET /api/content/<id>
- GET /api/stats
- GET /api/facets
- POST /api/chat {"message": "...", "context": []}

## MCP Server

For Claude Desktop integration, add to claude_desktop_config.json:
{
  "mcpServers": {
    "learning-library": {
      "command": "python",
      "args": ["/path/to/mcp_docent_server.py"]
    }
  }
}

MCP Tools: search_library, recommend_by_topic, get_learning_path, find_related_content, get_whats_new, get_content_excerpt

## Schema

Content entry structure:
{
  "id": "unique-id",
  "content_type": "video|paper|podcast|blog",
  "title": "Title",
  "url": "source-url",
  "facets": {
    "topics": ["ai-ml", "security"],
    "format": "tutorial|deep-dive|research-paper",
    "difficulty": "beginner|intermediate|advanced"
  },
  "summary": ["Key point 1", "Key point 2"],
  "sections": [{"title": "...", "description": "..."}]
}

## Rate Limits
- API: 100 requests/minute per IP
- For bulk access, use /library.json directly