Create evaluation agent
Register an external agent endpoint in Rippletide so you can upload knowledge, create test prompts, run evaluations, and read results. This does not build or host an agent for you.
Authorizations
Platform API key for authenticated account-level API access
Body
Register an external agent endpoint so Rippletide can run evaluations against it.
Human-readable name for the evaluation agent, such as Support Bot Eval.
"Support bot evaluation"
Deterministic seed for the evaluation setup. Use any integer; 123 is fine for manual calls.
123
Evaluation graph size hint. Use 100 for normal integrations.
100
Public HTTPS endpoint for your agent. Rippletide calls this endpoint when it runs tests.
"https://example.com/chat"
Optional request payload template for non-standard agent APIs. Omit this unless your endpoint does not accept the default message payload.
{
"payload": "{\"message\":\"[eval-question]\"}"
}Set to eval for public evaluation workflows.
eval "eval"
Response
Evaluation agent created
Evaluation agent registered in Rippletide. This record tells Rippletide what external agent endpoint to test and how to store evaluation results.
Human-readable name for this evaluation agent.
"Support bot evaluation"
Deterministic seed used by the evaluation setup. Use any integer; the CLI generates one automatically.
123
Evaluation graph size hint. Use 100 unless the Rippletide team recommends another value.
100
Unique agent identifier returned by Rippletide. Use this ID in later evaluation API calls.
"2f7e9c9c-9a2e-4e3a-b77f-6d9d1a7e3a11"
Public HTTPS endpoint for the agent Rippletide should evaluate.
"https://example.com/chat"
Optional request payload template used when Rippletide calls the agent endpoint. Omit this for standard chat-style APIs.
{
"payload": "{\"message\":\"[eval-question]\"}"
}Public integrations should use eval.
eval "eval"
Creation timestamp.
"2026-06-02T15:00:00Z"
Last update timestamp.
"2026-06-02T15:05:00Z"