F16.9 — LLM-friendly content endpoints

[jbourdin]

Summary

Provide lightweight, LLM-optimized endpoints that return clean text content (no HTML markup) suitable for retrieval-augmented generation (RAG) and AI agent consumption.

Details

Endpoints

• `GET /api/v1/llm/pages` — list published pages with plain-text content (Markdown stripped to text)
• `GET /api/v1/llm/pages/{slug}` — single page plain-text content
• `GET /api/v1/llm/events` — upcoming events with structured fields
• `GET /api/v1/llm/events/{id}` — single event detail
• `GET /api/v1/llm/archetypes` — published archetypes with descriptions
• `GET /api/v1/llm/decks/{shortTag}` — deck info with card list

Response format

• JSON with clean text fields (no HTML/Markdown)
• Compact payloads (no nested relations unless requested)
• Consistent structure for easy parsing by LLMs
• Include metadata: locale, lastModified, canonical URL

Authentication

• Public read endpoints (no OAuth required) for published content
• Rate-limited to prevent abuse
• Optional API key for higher rate limits

Dependencies

• Does NOT depend on F16.1 (OAuth) — uses simpler auth model
• Can be implemented independently of the full REST API

Acceptance Criteria

• Clean text endpoints for all public content types
• No HTML or Markdown in response text fields
• Rate limiting configured
• Response includes metadata (locale, lastModified, URL)
• Works without OAuth (public published content)