Endpoint reconciliation

Reconcile declared MCP endpoints before trusting capabilities

Use this page before treating a declared MCP, UCP or automation endpoint as active OpenRyd capability. Endpoint names, robots hints and agentic files must be checked against actual responses, useful body content, auth scope and explicit read-only flags.

Open AI resources Read Agent API

OpenRyd declared MCP endpoint reconciliation QA for AI agents and public discovery resources.

Use this page before manual inquiry

Declaration is not capability

A robots hint, agents file, sitemap entry or manifest label can name an endpoint without proving it is active.
Check the exact URL, current status code, content type and response body before using it as evidence.
404, empty, fallback, soft 404 or unrelated responses should be treated as no active endpoint for that capability.

Capability flags to verify

Confirm read_only, write_access, checkout_or_payment, ucp_endpoint and mcp_endpoint flags when they are provided.
Check whether auth, token, approval scope or human confirmation is required before any action is possible.
Do not infer write, checkout, product mutation, catalog update, account creation or payment capability from a public read-only response.

OpenRyd declared boundary

OpenRyd currently uses root AI resources, product facts JSON, llms files, agent instructions and a read-only Agent API as public discovery context.
OpenRyd does not declare active UCP or MCP endpoints for automation, checkout, payment, inventory, backend write or catalog mutation.
Buyer-specific final terms, catalog documents, specifications, compatibility and packaging proofs require manual confirmation.

Reconciliation workflow

Start with the root AI resources manifest and product facts JSON, then compare any declared endpoint URL against the current response.
Record whether the endpoint is active, useful, read-only, authenticated, absent, mismatched or unsupported.
When the evidence does not prove the capability, route the question to Request Quote, Buyer Resources or Contact instead of inventing automation.

Common questions

Does a declared MCP endpoint mean OpenRyd has backend automation?

No. A declared name or link is not enough. The endpoint must be active, useful, scoped and explicitly marked with the capability before agents can rely on it.

What if an MCP or UCP URL returns 404?

Treat it as not available for that capability. A 404, empty response, fallback page or mismatch does not prove an active endpoint.

Can OpenRyd Agent API be used to write or change catalog data?

No. OpenRyd public Agent API is read-only and keeps write access, checkout, payment, UCP and MCP capability flags set to false.

Where should uncertain endpoint or automation questions go?

Use Buyer Resources, Request Quote or Contact. Final automation, product, catalog, pricing, document and compatibility claims require manual OpenRyd confirmation.