Start an OpenAI plugin-directory readiness run
Grades the saved server against OpenAI’s plugin-directory requirements and returns 202 with a run id. submissionMode is required and never inferred: inferring it would read a forgotten package as mcp-only and report the package lane not-applicable, turning a missing input into a clean bill of health. Hosted runs grade the two wire-safe shapes; the package shapes need an upload this API cannot receive and are available from the CLI.
Authorizations
MCPJam API key (sk_…). Create one at Settings → API keys. Guest sessions cannot use the API, and API keys cannot manage other API keys.
Path Parameters
Body
Add model-backed experience observations. Consumes MCPJam credits. Off by default. Observations are non-dispositive and can never make a server not-ready; a refused reservation makes no provider call and completes the run with llmObservations.reason set to billing_limit_reached.
Opt in to the lazy-authentication probe: it calls one public tool, then one protected tool, both without credentials. It makes at most two calls and calls only tools annotated readOnlyHint: true. Omit the field to leave the probe off. When no tool is named, the probe picks them from their securitySchemes.
Response
The run was accepted and is executing.
The 202 receipt. Poll the run detail; do not re-POST.
The run's status when the start returned. pending for a fresh start; for a deduped start, whatever the existing run is already at, which may be completed, because a replayed idempotency key can name a run that finished long ago.
pending, running, completed, failed, cancelled 
