fix(dry-run): use sentinel failed run sample

Author: JerryNee

src/commands/test.test.ts

@@ -3025,7 +3025,7 @@ describe('runSteps', () => {
     expect(block.match(/error: /g)).toHaveLength(1);
   });
 
-  it('--run-id dry-run sample maps the failed step error and failure contributor flag', async () => {
+  it('--run-id run_failed_sample dry-run sample maps the failed step error and contributor flag', async () => {
     const out: string[] = [];
     const page = await runSteps(
       {
@@ -3034,7 +3034,7 @@ describe('runSteps', () => {
         debug: false,
         dryRun: true,
         testId: 'test_fe',
-        runId: 'run_dry',
+        runId: 'run_failed_sample',
       },
       {
         env: {} as NodeJS.ProcessEnv,

src/lib/dry-run/samples.test.ts

@@ -424,9 +424,25 @@ describe('findSample', () => {
     expect(e).toBeUndefined();
   });
 
-  it('GET /runs/{runId} sample includes a failed run-scoped step', () => {
+  // defect-2 fix: getRun sample must return the passed shape (first-match-wins
+  // in findSample). Prior to fix, a duplicate failed-shape entry appeared
+  // before the passed-shape entry; `test wait --dry-run` always resolved to
+  // status: "failed", giving agents the wrong happy-path canned response.
+  it('GET /runs/{runId} resolves to the passed-shape getRun (not the failed shape)', () => {
     const e = findSample('GET', 'https://api.testsprite.com/api/cli/v1/runs/run_xyz');
     expect(e?.operationId).toBe('getRun');
+    const body = e?.body() as {
+      status: string;
+      runId: string;
+      stepSummary: { failedCount: number };
+    };
+    expect(body.status).toBe('passed');
+    expect(body.stepSummary.failedCount).toBe(0);
+  });
+
+  it('GET /runs/run_failed_sample resolves to the sentinel failed run-scoped step sample', () => {
+    const e = findSample('GET', 'https://api.testsprite.com/api/cli/v1/runs/run_failed_sample');
+    expect(e?.operationId).toBe('getRun');
     const body = e?.body() as {
       status: string;
       runId: string;
@@ -441,6 +457,7 @@ describe('findSample', () => {
         error: string | null;
       }>;
     };
+    expect(body.runId).toBe('run_failed_sample');
     expect(body.status).toBe('failed');
     expect(body.failedStepIndex).toBe(3);
     expect(body.failureKind).toBe('assertion');

src/lib/dry-run/samples.ts

@@ -52,6 +52,10 @@ const SAMPLE_TEST_ID_FAILED = 'test_8f2a4d10';
 const SAMPLE_TEST_ID_PASSED = 'test_3a91bb02';
 const SAMPLE_TEST_ID_BLOCKED = 'test_blocked_4f7a';
 export const SAMPLE_RUN_ID = 'run_abc';
+// Documented sentinel for `test steps --run-id run_failed_sample --dry-run`:
+// keeps wait flows on the default passed sample while still demonstrating a
+// run-scoped failed step offline.
+const SAMPLE_FAILED_RUN_ID = 'run_failed_sample';
 // M3.4 rerun dry-run sample IDs
 const SAMPLE_RERUN_ID_BE_NAMED = 'run_rerun_be_named';
 const SAMPLE_RERUN_ID_BE_PRODUCER = 'run_rerun_be_producer';
@@ -347,6 +351,118 @@ const failureSummary: CliFailureSummary = {
   recommendedFixTarget: failureContext.failure.recommendedFixTarget,
 };
 
+const passedRunSample: RunResponse = {
+  runId: SAMPLE_RUN_ID,
+  testId: SAMPLE_TEST_ID_PASSED,
+  projectId: SAMPLE_PROJECT_ID,
+  userId: SAMPLE_USER_ID,
+  status: 'passed',
+  source: 'cli',
+  createdAt: '2026-05-15T19:32:00.000Z',
+  startedAt: '2026-05-15T19:32:05.000Z',
+  finishedAt: '2026-05-15T19:34:00.000Z',
+  codeVersion: 'v1',
+  targetUrl: SAMPLE_TARGET_URL,
+  createdFrom: null,
+  failedStepIndex: null,
+  failureKind: null,
+  error: null,
+  videoUrl: null,
+  stepSummary: {
+    total: 8,
+    completed: 8,
+    passedCount: 8,
+    failedCount: 0,
+  },
+  // Representative per-run steps so `test steps --run-id <id> --dry-run`
+  // demonstrates real output instead of an empty list (the generic
+  // `/runs/{runId}` sample is also used by `test wait`, which ignores steps).
+  steps: [
+    {
+      stepIndex: '0001',
+      type: 'action',
+      action: 'navigate',
+      status: 'passed',
+      description: 'Open the target URL',
+      error: null,
+      screenshotUrl: null,
+      htmlSnapshotUrl: null,
+      createdAt: '2026-05-15T19:32:10.000Z',
+    },
+    {
+      stepIndex: '0002',
+      type: 'assertion',
+      action: 'assert_visible',
+      status: 'passed',
+      description: 'Dashboard heading is visible',
+      error: null,
+      screenshotUrl: null,
+      htmlSnapshotUrl: null,
+      createdAt: '2026-05-15T19:32:20.000Z',
+    },
+  ],
+};
+
+const failedRunSample: RunResponse = {
+  runId: SAMPLE_FAILED_RUN_ID,
+  testId: SAMPLE_TEST_ID_FAILED,
+  projectId: SAMPLE_PROJECT_ID,
+  userId: SAMPLE_USER_ID,
+  status: 'failed',
+  source: 'cli',
+  createdAt: '2026-05-15T19:32:00.000Z',
+  startedAt: '2026-05-15T19:32:05.000Z',
+  finishedAt: '2026-05-15T19:34:00.000Z',
+  codeVersion: 'v1',
+  targetUrl: SAMPLE_TARGET_URL,
+  createdFrom: null,
+  failedStepIndex: 3,
+  failureKind: 'assertion',
+  error: 'Expected billing status badge to be visible, but it was not found.',
+  videoUrl: null,
+  stepSummary: {
+    total: 3,
+    completed: 3,
+    passedCount: 2,
+    failedCount: 1,
+  },
+  steps: [
+    {
+      stepIndex: '0001',
+      type: 'action',
+      action: 'navigate',
+      status: 'passed',
+      description: 'Open the target URL',
+      error: null,
+      screenshotUrl: null,
+      htmlSnapshotUrl: null,
+      createdAt: '2026-05-15T19:32:10.000Z',
+    },
+    {
+      stepIndex: '0002',
+      type: 'assertion',
+      action: 'assert_visible',
+      status: 'passed',
+      description: 'Dashboard heading is visible',
+      error: null,
+      screenshotUrl: null,
+      htmlSnapshotUrl: null,
+      createdAt: '2026-05-15T19:32:20.000Z',
+    },
+    {
+      stepIndex: '0003',
+      type: 'assertion',
+      action: 'assert_visible',
+      status: 'failed',
+      description: 'Billing status badge is visible',
+      error: 'Expected billing status badge to be visible, but it was not found.',
+      screenshotUrl: null,
+      htmlSnapshotUrl: null,
+      createdAt: '2026-05-15T19:32:30.000Z',
+    },
+  ],
+};
+
 /**
  * Dry-run sample lookup keyed by OpenAPI operationId. Order matters in
  * {@link findSample}: more specific patterns must precede their generic
@@ -702,71 +818,12 @@ const ENTRIES: DryRunSampleEntry[] = [
     },
   } satisfies BatchRerunResponse),
   // M3.3 piece-3 — GET /runs/{runId} (live status / long-poll).
-  // Use a terminal failed row with a concrete failed step so
-  // `test steps --run-id <id> --dry-run` demonstrates the run-scoped
-  // error + failedStepIndex mapping without needing live credentials.
-  entry('getRun', 'GET', '/runs/{runId}', {
-    runId: SAMPLE_RUN_ID,
-    testId: SAMPLE_TEST_ID_FAILED,
-    projectId: SAMPLE_PROJECT_ID,
-    userId: SAMPLE_USER_ID,
-    status: 'failed',
-    source: 'cli',
-    createdAt: '2026-05-15T19:32:00.000Z',
-    startedAt: '2026-05-15T19:32:05.000Z',
-    finishedAt: '2026-05-15T19:34:00.000Z',
-    codeVersion: 'v1',
-    targetUrl: SAMPLE_TARGET_URL,
-    createdFrom: null,
-    failedStepIndex: 3,
-    failureKind: 'assertion',
-    error: 'Expected billing status badge to be visible, but it was not found.',
-    videoUrl: null,
-    stepSummary: {
-      total: 3,
-      completed: 3,
-      passedCount: 2,
-      failedCount: 1,
-    },
-    // Representative per-run steps so `test steps --run-id <id> --dry-run`
-    // demonstrates real output instead of an empty list (the generic
-    // `/runs/{runId}` sample is also safe for wait flows, which ignore steps).
-    steps: [
-      {
-        stepIndex: '0001',
-        type: 'action',
-        action: 'navigate',
-        status: 'passed',
-        description: 'Open the target URL',
-        error: null,
-        screenshotUrl: null,
-        htmlSnapshotUrl: null,
-        createdAt: '2026-05-15T19:32:10.000Z',
-      },
-      {
-        stepIndex: '0002',
-        type: 'assertion',
-        action: 'assert_visible',
-        status: 'passed',
-        description: 'Dashboard heading is visible',
-        error: null,
-        screenshotUrl: null,
-        htmlSnapshotUrl: null,
-        createdAt: '2026-05-15T19:32:20.000Z',
-      },
-      {
-        stepIndex: '0003',
-        type: 'assertion',
-        action: 'assert_visible',
-        status: 'failed',
-        description: 'Billing status badge is visible',
-        error: 'Expected billing status badge to be visible, but it was not found.',
-        screenshotUrl: null,
-        htmlSnapshotUrl: null,
-        createdAt: '2026-05-15T19:32:30.000Z',
-      },
-    ],
-  } satisfies RunResponse),
+  // A terminal `passed` row is the most useful dry-run shape: agents see
+  // what a completed run looks like, and `--wait` terminates immediately.
+  // fix(2026-05-21): a duplicate failed-shape entry that appeared before
+  // this entry was removed; findSample first-match-wins was always
+  // returning status: "failed" for `test wait --dry-run`.
+  entry('getRun', 'GET', '/runs/{runId}', passedRunSample),
   // DEV-331 piece 3 — POST /runs/{runId}/cancel. Method-guarded in
   // `findSample` (POST vs `getRun`'s GET), so this can't be shadowed by the
   // broader `/runs/{runId}` pattern above despite sharing its path prefix.
@@ -848,11 +905,15 @@ export function findSample(
   const pathOnly = extractPath(url);
   for (const e of ENTRIES) {
     if (e.method === upper && e.pattern.test(pathOnly)) {
+      const body =
+        e.operationId === 'getRun' && pathOnly === `/runs/${SAMPLE_FAILED_RUN_ID}`
+          ? failedRunSample
+          : e.body(requestBody);
       // Rebind body so callers get the resolved value, not the factory.
       // We return a new object with `body` already applied so downstream
       // code can keep calling `e.body` as-before (no API break for tests
       // that call `findSample` directly).
-      return { ...e, body: () => e.body(requestBody) };
+      return { ...e, body: () => body };
     }
   }
   return undefined;