A public surface for agents. Human authority for consequential actions.
RigorLoop publishes clear discovery resources and a machine-readable contract. Agents can find, read, and validate public information today. Submission, funding, expert selection, and review remain deliberately bounded.
Canonical public surfaces
Start from these locations. The OpenAPI contract—not an inferred URL or interface label—defines current API capability.
- API namespace
https://rigorloop.com/api/v1- Capability status
- https://rigorloop.com/api/v1/capabilities
- Funding options
- https://rigorloop.com/api/v1/funding/options
- OpenAPI contract
- https://rigorloop.com/openapi.json
- Machine guide
- https://rigorloop.com/llms.txt
The contract determines which operations are currently available.
Live read, validation, submission, review, and payment boundaries.
Stripe is human-controlled; Base and USDC are not live yet.
Schemas, authentication, responses, and current operation status.
A concise index of public resources and agent boundaries.
A safe integration flow
Machine-readable does not mean permissionless. Capability and authority stay explicit at every step.
Discover the public surface
Start with llms.txt and the sitemap. Use stable URLs instead of guessing from interactive controls.
Read the contract
Treat openapi.json as authoritative for available operations, schemas, authentication, and errors.
Validate before calling
Validate request and response data against the published schema. Do not infer undocumented write methods.
Hand consequential work to a person
A signed-in human controller creates or funds a claim, chooses an expert, and makes payment decisions.
The human review boundary is structural.
Agents may help a controller prepare or inspect work, but they are never eligible to be a Prover. A signed review always comes from the vetted person whose identity and expertise are attached to it.