test(wp7-02): require independent mixed evidence review
Dada P0-A isolated Windows CI / validate-and-package (push) Failing after 2m8s

This commit is contained in:
suyx
2026-08-04 17:33:51 +08:00
parent 63de0917a0
commit 382d50058c
3 changed files with 64 additions and 1 deletions
@@ -256,11 +256,16 @@ test("TDD-WP7-EXT-001 assembles independent complete evidence without retaining
const evidence = assembleControlledModelEvidence({
deterministicState,
modelConfig: models[0],
realExecution: { calls, planned_real_calls: 5, real_calls: 5, status: "passed" },
realExecution: {
attempts: calls.map((call, index) => ({ attempt_no: 1, http_status: 200, scenario_id: call.scenario_id, status: "passed" })),
calls, maximum_real_calls: 6, planned_real_calls: 5, real_calls: 5, status: "passed",
},
runId: "wp7-02-assembly-test",
});
assert.equal(evidence.status, "passed");
assert.equal(evidence.manual_review.status, "pending");
assert.equal(evidence.external_calls.attempts.length, 5);
assert.equal(evidence.external_calls.maximum_real_calls, 6);
assert.equal(evidence.matrix.error_scenarios.length, 9);
assert.doesNotMatch(JSON.stringify(evidence), /iVBOR|image_bytes|raw_prompt|authorization/i);
const execution = buildDeterministicExecutionEvidence(models[0].model_id, "wp7-02-assembly-test");