# mythoi.health — the stories

A reference database of the world's myths. Traditions shelved as
collections; each work read in episodes with canonical references;
Thompson's Motif-Index as the address layer (every motif is a page).
Every claim carries a receipt tracing it to the text it stands on.

The triad: seba.health (the person) · logoi.health (the words) ·
mythoi.health (the stories).

## Pages
- /motifs — the Thompson chapter lattice, A to Z; /motif/<tmi-id> is a motif's page (e.g. /motif/A1010)
- /traditions and /tradition/<slug> — the shelves
- /tales and /tale/<slug> — the episode reader (license-gated text)
- /receipt/<slug> — receipts: sources consulted, checksums, what a claim does not prove
- /rhapsode — Ask the Rhapsode, the song-stitcher (AI over tool results only)

## Agent channel
- GET /api/v1/tools — list the tools and their JSON schemas
- GET/POST /api/v1/tools/<tool> — run a tool (motif_lookup, motif_neighbors, attestation_search, tradition_profile, episode_lookup, receipt_get). GET maps query params to input; POST takes JSON input. Every result is an envelope with source_ids + receipt_url. motif_neighbors walks the knowledge graph: Thompson's tradition citations, the index hierarchy, his cross-references, and ATU tale-type links.
- POST /api/mcp — Model Context Protocol (streamable HTTP, JSON responses): initialize, tools/list, tools/call over the same five tools, receipts in every result.
- POST /api/v1/ask — { question } → { answer, thread } (the Rhapsode; opens when the model key is set).

## License posture
Public-domain and open sources are served with attribution; NC and
copyrighted material is pointer-only. Per-item license fields ride on
every source row and every episode.