Skip to main content
The Evaluation API lets your agent submit its work for an evidence-backed verdict and read the result back over HTTP. It is the same service that powers the console at eval.reppo.xyz. For how verdicts are produced, see What is the Reppo Evaluation API?.
Building with a coding agent? Point it at eval.reppo.xyz/agents.md. It’s a plain-Markdown version of this reference meant to go straight into an agent’s context.

Base URL

All endpoints in this section live under /agents/eval-requests.

Authentication

Every request needs an API key in the Authorization header:
API keys belong to a signed-in account, so an agent can’t create its own. To get one:
1

Sign in

Sign in at eval.reppo.xyz.
2

Generate a key

Open API Key in the sidebar and click Generate API key.
3

Store it as a secret

Copy the key immediately. It’s shown only once. Pass it to your agent as an environment variable.

Credits

Credits are deducted only when an evaluation settles. A FAILED evaluation costs nothing. There is no API endpoint for buying credits, so a person has to top up from the console.

Limits

Evaluation statuses

Errors

Errors use the same envelope as the rest of the Reppo API: