Version 1.1 · public · read-only

Developers

Dhamma data exposes bounded retrieval over its public Pāli/Theravāda corpus, dictionary, concordance, parallel references, passage comparison, evidence analysis, and library. MCP and REST share the same validation, provenance, rights, and privacy boundary.

No account or key is required. Public operations retrieve evidence and do not write to the corpus, personal collections, or unpublished administrative research.

Open interactive developer documentation →

MCP

Endpoint: https://dhamma.fly.dev/mcp

Transport: stateless Streamable HTTP. Authentication: none. Tool discovery returns the current schemas and remains the authority for accepted fields and bounds.

{
  "mcpServers": {
    "dhamma_data": {
      "type": "http",
      "url": "https://dhamma.fly.dev/mcp"
    }
  }
}

Public tool surface

Search and read public corpus passages with stable identifiers, citations, textual layers, and source links.
Inspect passage context, translations, commentary relationships, parallel references, tags, and corpus details.
Browse the Pāli/Theravāda corpus, compare selected held readings, trace concepts across declared strata, and analyze bounded caller-supplied evidence sets.
Look up Pāli terms and related forms, build bounded concordances, and search or read public library material.
Use the standard search/fetch compatibility pair when a client requires opaque retrieval identifiers.

REST API

Base URL: https://dhamma.fly.dev/api/v1

Discovery: https://dhamma.fly.dev/api/v1 · OpenAPI 3.1: https://dhamma.fly.dev/api/v1/openapi.json

GET endpoints cover corpus search, passages and context, dictionary lookup, concordance, browsing, library retrieval, sources, and coverage. Bounded JSON POST endpoints accept structured witness comparison, concept tracing, and evidence analysis inputs.

curl https://dhamma.fly.dev/api/v1
curl https://dhamma.fly.dev/api/v1/openapi.json
curl "https://dhamma.fly.dev/api/v1/search?q=sati&mode=stem&limit=5"

Results and source rules

Applicable results include a stable identifier, human citation, absolute Dhamma data URL, textual layer, and edition or upstream source.
Attributed translations and articles include the named translator or author and item-specific copyright, licence, and source metadata where available.
Pagination, cursors, availability, and truncation fields must be respected. A retrieval or ranking result is discovery evidence, not proof that terms or traditions are equivalent.
Item-specific rights metadata takes priority over catalogue fallbacks. API access does not create a new licence to underlying text.

Errors, privacy, and bounds

Invalid inputs use machine-readable client errors; unknown records return not found; rate limits use HTTP 429 and Retry-After. Temporary bounded retrieval failures must not be interpreted as completed searches or evidence of absence.

Public request and response content is not retained as a user history. Private translations, bookmarks, notes, research dossiers, personal collections, and unpublished research are outside the public gateway.

Respect declared maxima, pagination, backoff, attribution, and source-specific rights. Bulk corpus reconstruction, limit evasion, and stripping provenance are not permitted.