Rerun Call Executions
Rerun call executions within a test execution. Choose eval_only to re-evaluate existing data or call_and_eval to re-run calls. Returns success and failure counts per call.
https://api.futureagi.com/simulate/test-executions/{test_execution_id}/rerun-calls/ Authentication
Path parameters
The test execution ID. Must be in a terminal state (completed, failed, or cancelled).
Request body
The type of rerun to perform. Use eval_only to re-evaluate existing call data without re-executing the actual calls — this is useful when you have updated your evaluation configurations and want to see updated scores without the cost of re-running calls. Use call_and_eval to fully re-execute the calls and then evaluate the new results — this produces fresh conversations and is useful when you have modified the agent under test. Note that text agents only support eval_only reruns; attempting call_and_eval on a text agent will return a 400 error.
An array of call execution UUIDs to rerun. Required when select_all is false or not provided. Each ID must correspond to a valid call execution within the specified test execution. If a provided ID does not exist or does not belong to the test execution, it will appear in the failed_reruns array of the response.
When set to true, all call executions within the test execution will be rerun, and the call_execution_ids field is ignored. Defaults to false. You must provide either select_all: true or a non-empty call_execution_ids array — the request will fail with a 400 error if neither is specified.
Response
200 OKA human-readable confirmation message indicating that the rerun has been initiated. The actual rerun processing happens asynchronously after this response is returned.
The UUID of the test execution that the rerun was initiated for, echoed back for confirmation and reference.
The type of rerun that was requested, either eval_only or call_and_eval. Echoed back from the request for confirmation.
The total number of call executions that were processed by the rerun request. This includes both successful and failed reruns.
An array of call execution UUIDs that were successfully queued for rerun. These calls will be re-executed or re-evaluated asynchronously.
An array of objects describing call executions that could not be rerun. Each object contains a call_execution_id (the UUID of the failed call) and an error (a human-readable description of why the rerun failed, such as the call being in an incompatible state).
The number of call executions that were successfully queued for rerun. Equal to the length of the successful_reruns array.
The number of call executions that failed to be queued for rerun. Equal to the length of the failed_reruns array.
Errors
The rerun request could not be processed. This error occurs when: the rerun_type field is missing or contains an invalid value; neither call_execution_ids nor select_all was provided; the test execution is still in an active state (pending, running, or cancelling) and cannot accept reruns; or a call_and_eval rerun was requested for a text agent, which only supports eval_only reruns. Check the error message in the response body for specific details on which validation failed.
The request could not be authenticated. Verify that both X-Api-Key and X-Secret-Key headers are present and contain valid, non-expired credentials. Ensure the API key has access to the workspace that owns this test execution.
Test execution not found.
Unexpected server error.