Skip to main content
POST /api/v2/agents/{agentId}/creative-conformance Runs the AdCP creative storyboards selected from the Agent’s own get_adcp_capabilities response. The test calls only the Agent’s attested production HTTPS endpoint and does not create a creative or spend money. Only a directly authenticated registration administrator for the owning organization can start a run. The Agent must be a claimed CREATIVE Agent with an attested production endpoint. Apostra allows one run at a time for an Agent and accepts another run five minutes after recorded evidence. The endpoint’s own timeout or a malformed capabilities response can stop a run before it records evidence. The runner calls only a public HTTPS endpoint that passes Apostra’s outbound endpoint guard. Each storyboard step has a 10-second timeout, the whole run has a 90-second timeout, and each response is limited to 1 MB.

Request

Parameters

Response

Each step is PASSED, FAILED, or SKIPPED. Apostra stores recorded results as evidence for the production revision that was tested. Read the Agent again with GET /api/v2/agents/{agentId} to receive that revision’s latest result in creativeConformance. A result is conformance evidence, not creative-agent certification; creative certification is not available yet. Passed evidence expires after seven days; failed evidence remains available. Deploying a new production revision hides the prior revision’s result until that revision is tested.

Errors

  • 400 VALIDATION_ERROR - the Agent has no attested production endpoint, does not declare the creative protocol, the endpoint is unsafe, or the runner did not execute an assertion.
  • 403 ACCESS_DENIED - the caller is not a directly authenticated registration administrator for the owning organization.
  • 404 NOT_FOUND - the Agent is not a claimed Creative Agent owned by the organization.
  • 409 CONFLICT - a run is already in progress or the five-minute rate limit has not elapsed.
See Errors for the full error contract.

Agent tasks

All agent operations

Storefront agents

Agent model and page guidance