Canonical answer reference
Embedded Semantics answers
Direct, citable answers generated from the same catalog as the JSON, Markdown, OpenAPI, and MCP discovery records.
What is Embedded Semantics?
Embedded Semantics is a governed expression-resolution service. It maps an exact expression to a published ConceptCode only when the resolver returns a structurally valid resolved outcome.
How do AI agents connect?
Agents can use the direct JSON API or the stateless MCP endpoint. Start with GET /api/v1/agent/connect for API directions or POST /mcp for MCP discovery.
When may an agent accept a ConceptCode?
An agent may accept exactly one ConceptCode only from a structurally valid resolved response. Unknown, ambiguous, malformed, request-error, service-error, and transport-error outcomes accept none.
How do machines discover the public APIs?
Read /.well-known/api-catalog for the RFC 9727 Linkset catalog, /openapi.json for the direct API contract, or /mcp-server.json for the deployment-specific MCP manifest.
Are machine-readable representations byte-stable?
Owned JSON, Markdown, Linkset, and discovery representations use deterministic UTF-8 bytes with strong content-derived validators. Nonce-bearing HTML is private and no-store instead.
How can an agent verify a discovery representation?
Read /.well-known/embedded-semantics-discovery.json, select a route from its representation receipt, and compare the returned Content-Location, media type, strong ETag, and Repr-Digest with the exact response bytes. Verify deployment identity through an independently authenticated release channel.