Run Under Test
Record run ID, objective, agent, model, output link, changed files, and verifier.
Check objective match, scope, behavior, regression, security, cost, handoff, and revenue-proof evidence before accepting agent work.
Record run ID, objective, agent, model, output link, changed files, and verifier.
Require direct behavior proof, relevant regression checks, security/privacy checks, and handoff clarity.
Hold when proof is missing, indirect, stale, out of scope, or not understandable without chat history.
Revenue is proven only by checkout, receipt, payout, or seller-dashboard evidence. Interest, stars, views, downloads, and issue text are not revenue.