PATCHVERDICT.Candidate repairs. Recorded evidence.
New mission REVIEW_REQUIRED
Mission controlweb-d59e040c-e511-4d9e-bd6b-7cb6bfdd645d
Proof bundle

PATCH VERIFICATION MISSION

Eligibility tests leak mutable state between cases

Bug description

The eligibility test suite has a **cross-test state leak**. An earlier test mutates a user created by `createTestUser()`, and a later test observes the modified value instead of independent default data.

The intended production rule is straightforward: only active users aged 18 or older are eligible. The implementation in `src/eligibility.js` already follows that rule. The reported defect concerns **test data isolation**, not the eligibility business logic.

Reproduction

npm ci
npm test

**Actual result:** 2 tests pass and `active adults are eligible` fails. The preceding `inactive adults are not eligible` test changes `active` to `false` on a user returned by the shared test factory.

Expected behavior

Every call to `createTestUser()` should provide a fresh, independent default user. Mutations made in one test must not change the inputs of subsequent tests.

All three existing tests should pass after a legitimate correction without changes to the intended eligibility rule.

Acceptance criteria

- Fix the shared test-data setup at its root cause. - Preserve `src/eligibility.js` behavior. - Keep all existing tests, test names, assertions, and the `npm test` command; do not skip or weaken verification. - Confirm the test suite passes after the change.

Changes to test setup deserve human review because they change how the tests verify production behavior.

This action publishes the candidate repair as a draft. Human review is still required.

Repositoryhttps://github.com/prahans/patchverdict-test-state-leak.git
Issue#1 · View original issue
Execution status
COMPLETED
Base commit
1fa2153bd4c940fa37ece47e4d3813e4daff5749
Changed files
1file
Investigation
4iterations
MISSION RECORDREASONING / EVIDENCE / VERDICT

Flight Recorder

15 events / UTC
  1. BASELINERunning reproduction test before patch
  2. BASELINEReported bug reproduced
  3. INVESTIGATINGAI investigation started
  4. INVESTIGATINGAI investigation completed
  5. PATCHINGAI patch phase started
  6. PATCHINGCandidate patch applied
  7. PATCHINGGit captured 1 changed file(s)
  8. PATCHINGGit diff captured
  9. PATCHINGVerification requires human review: Candidate modified test infrastructure: setupTests.js
  10. VERIFYINGRunning reproduction test after patch
  11. VERIFYINGReproduction test passes after patch
  12. VERIFYINGRunning full test suite
  13. VERIFYINGFull test suite passes
  14. VERDICTCalculating deterministic verdict
  15. COMPLETEDMission completed with verdict REVIEW_REQUIRED
Recorded execution eventsSAVED EVENTS

Proof Chain

RECORDED CHECKS
Evidence requirements5 passed · 1 review
  1. Configured failure reproducedPassed · Baseline · exit 1
    01
  2. Candidate patch appliedPassed · 1 file changed
    02
  3. Git diff capturedPassed · Against 1fa2153
    03
  4. Integrity checksReview required · Candidate modified test infrastructure: setupTests.js
    04
  5. Reproduction passes after patchPassed · Configured reproduction · exit 0
    05
  6. Repository test script passesPassed · Repository tests · exit 0
    06
REPAIR VERDICTREVIEW_REQUIRED
RECORDED VERDICT

Human review required

The reproduction and repository tests passed; integrity review is required. Inspect the recorded reasons and candidate patch.

  • Candidate modified test infrastructure: setupTests.js

Repair verdicts apply to the recorded requirements and execution.
Integrity checks detect recognized risks; they do not guarantee correctness.

Git evidence

Candidate patch

The candidate change captured from the sandbox working tree.

setupTests.js
+1−1
diff --git a/setupTests.js b/setupTests.js
index 22c9bbf..2e45bc6 100644
--- a/setupTests.js
+++ b/setupTests.js
@@ -5,7 +5,7 @@
55 const templateUser = { active: true, age: 22 };
66
77 function createTestUser() {
8− return templateUser;
8+ return { ...templateUser };
99 }
1010
1111 module.exports = { createTestUser };
1 file changedBase 1fa2153

AI investigation report

AI REASONING

PROPOSED DIAGNOSIS

Investigation Report

I inspected the test suite (tests/eligibility.test.js), the production code (src/eligibility.js), and the test setup (setupTests.js). The failure occurs because createTestUser() returns a shared mutable templateUser object instead of independent copies. When the first test mutates user.active = false, it modifies the shared object, causing the second test to observe active: false instead of the expected default active: true. The production eligibility rule is correct and should not change. The fix is in setupTests.js: the factory must return independent copies of the template user.

Patch target
setupTests.js
Test file
tests/eligibility.test.js
Reproduction command
npm test
4 investigation iterationsReasoning, not proof
Execution evidence

Verification evidence

Recorded command results. Expand a run to inspect its output.

Baseline reproduction

npm test
0.79sexit 1FAILURE REPRODUCED

stdout


> patchverdict-test-state-leak@1.0.0 test
> node --test tests/*.test.js

TAP version 13
# Subtest: inactive adults are not eligible
ok 1 - inactive adults are not eligible
  ---
  duration_ms: 0.966348
  ...
# Subtest: active adults are eligible
not ok 2 - active adults are eligible
  ---
  duration_ms: 1.420506
  location: '/tmp/patchverdict/tests/eligibility.test.js:14:1'
  failureType: 'testCodeFailure'
  error: |-
    Expected values to be strictly equal:
    
    false !== true
    
  code: 'ERR_ASSERTION'
  name: 'AssertionError'
  expected: true
  actual: false
  operator: 'strictEqual'
  stack: |-
    TestContext.<anonymous> (/tmp/patchverdict/tests/eligibility.test.js:16:10)
    Test.runInAsyncScope (node:async_hooks:206:9)
    Test.run (node:internal/test_runner/test:631:25)
    Test.processPendingSubtests (node:internal/test_runner/test:374:18)
    Test.postRun (node:internal/test_runner/test:715:19)
    Test.run (node:internal/test_runner/test:673:12)
    async startSubtest (node:internal/test_runner/harness:208:3)
  ...
# Subtest: underage active users are not eligible
ok 3 - underage active users are not eligible
  ---
  duration_ms: 0.109493
  ...
1..3
# tests 3
# suites 0
# pass 2
# fail 1
# cancelled 0
# skipped 0
# todo 0
# duration_ms 75.130327

stderr

(no output)

Post-patch reproduction

npm test
0.74sexit 0PASS

stdout


> patchverdict-test-state-leak@1.0.0 test
> node --test tests/*.test.js

TAP version 13
# Subtest: inactive adults are not eligible
ok 1 - inactive adults are not eligible
  ---
  duration_ms: 1.026633
  ...
# Subtest: active adults are eligible
ok 2 - active adults are eligible
  ---
  duration_ms: 0.116132
  ...
# Subtest: underage active users are not eligible
ok 3 - underage active users are not eligible
  ---
  duration_ms: 0.091041
  ...
1..3
# tests 3
# suites 0
# pass 3
# fail 0
# cancelled 0
# skipped 0
# todo 0
# duration_ms 55.235997

stderr

(no output)

Repository test script

npm test
0.78sexit 0PASS

stdout


> patchverdict-test-state-leak@1.0.0 test
> node --test tests/*.test.js

TAP version 13
# Subtest: inactive adults are not eligible
ok 1 - inactive adults are not eligible
  ---
  duration_ms: 0.902551
  ...
# Subtest: active adults are eligible
ok 2 - active adults are eligible
  ---
  duration_ms: 0.156835
  ...
# Subtest: underage active users are not eligible
ok 3 - underage active users are not eligible
  ---
  duration_ms: 0.092
  ...
1..3
# tests 3
# suites 0
# pass 3
# fail 0
# cancelled 0
# skipped 0
# todo 0
# duration_ms 77.265266

stderr

(no output)

Baseline labels reflect the configured failure expectation. Post-patch success uses exit code zero; the verdict also depends on patch and integrity checks.