How can software and AI systems retrieve SaSame knowledge?
OpenAPI direction, reference, examples, schemas and authentication boundaries.
- Collection
- api
- Updated
- 2026-07-30
- min read
- 2
- Version
- v1
Use the read-only Knowledge API, search endpoint or MCP retrieval tools, all backed by the same document and relation model.
Knowledge API
Documents and relations return versioned machine-readable records with locale, lifecycle status and provenance fields.
MCP
Public knowledge tools are read-only and expose no owner shell, credentials or administration.
Authentication
Public reads can be anonymous; publishing and protected datasets require scoped authentication and receipts.
Concrete endpoints
Read-only surfaces available today: /api/knowledge and /api/search (the Knowledge API and search endpoint), /openapi.json (the machine-readable API contract), /llms.txt (a plain-text index for language-model retrieval), /rss.xml (recently updated records) and /sitemap.xml. All are unauthenticated and read-only; nothing under /api/admin/* is reachable without a session, and the admin surface never appears in navigation, sitemap or search.
Examples
2- 01
GET /api/knowledge?slug=research returns a canonical document and related pages.
- 02
GET /openapi.json returns the exact machine-readable contract for the Knowledge API, so a client can validate its integration against the same schema this site itself uses.