How do I install the E2e Fixer skill?

Run npx skillmds@latest add conceptadev/e2e-fixer in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is E2e Fixer safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with E2e Fixer?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is E2e Fixer free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published E2e Fixer?

conceptadev (@conceptadev) published this skill. Their other Agent Skills are listed on their SkillMD profile.

E2e Fixer by conceptadev — Agent Skill

Name: E2e Fixer
Author: conceptadev

E2e Fixer

Diagnose and fix failing or flaky tests in this monorepo, with a bias for e2e (*.e2e-spec.ts). Use when tests fail after a change, when a suite passes alone but fails in the full run, on fixture/bootstrap drift, barrel-registration collisions, teardown/open-handle leaks, or runner/config drift. Triggers on "fix the tests", "e2e failing", "flaky test", "tests pass alone but not together".

conceptadev 7afbfdf 3.2 KB Updated Sep 17, 2026

File contents

E2E Fixer

E2E is the default test tier here (*.e2e-spec.ts, real Nest app + supertest + SQLite). Most "failures" are environment/isolation, not logic — separate the two before editing source.

First: is it mine, pre-existing, or flaky?

Run the failing suite in isolation: corepack yarn vitest run --project e2e-packages "<suite-path>". Passes alone but fails in the full run → see the known rotating-404 flake in CHANGELOG.md before digging.
Establish a baseline with git stash (keep node_modules) and re-run the suite. Same failure on clean HEAD → pre-existing, not your regression. Say so explicitly; do not "fix" pre-existing breakage silently.

Known failure classes and fixes

Rotating full-run failures = host memory pressure, not a code bug. Symptoms vary (unexpected 404, an expected 401 arriving as 404, Parse Error: Expected HTTP/) with a different victim suite each run. Tell: the failing run takes ~20x longer than a passing one (216s vs 10.8s) and the real failures are 30s timeouts with no HTTP response. Check free RAM / pageouts before touching any test. Diagnosis and evidence in CHANGELOG.md "Intermittent e2e failures". Fix the host (--maxWorkers=2, free memory) — never retry-loop, skip, or weaken assertions.
Barrel registration collisions. Importing a domains/*/index barrel registers @CommandHandler/ @QueryHandler in global Reflect metadata and breaks later Nest apps in the same process. Never import a barrel in an e2e file that boots an app. (The barrel-last sequencer is gone; no barrel-only specs exist.)
Teardown / open handles. A hanging suite process + Parse Error: Expected HTTP/ from supertest = an app wasn't closed. Ensure await app.close() in afterEach/afterAll. forceExit is off on purpose — a leak hangs and gets reported instead of being masked.
Fixture drift (v7→v8). Symptoms: 500s, Class extends value undefined, x is not a function. Fix the fixture to the v8 pattern — RepositoryModule.forFeature for every entity, no duplicate TypeOrmModule.forRoot, correct extras (user/otp/role/federated/invitation). Reuse packages/rockets-server-auth/src/__fixtures__/ and __e2e__/helpers/.
Module-resolution after a bump. Cannot find module '@concepta/.../subpath' → see upstream-migrator. Vitest resolves the package exports maps natively; if a subpath fails, the exports map itself is wrong.

Rules

New tests must be *.e2e-spec.ts unless a unit test is specifically justified. Coverage target ≥ 80% via yarn test:e2e:cov.
Fix the test setup or the real bug — never weaken an assertion or add any to make red go green.
Report the final tally honestly (passed/failed/skipped) and label each remaining failure as fixed, pre-existing, or flaky.

---name: e2e-fixerdescription: Diagnose and fix failing or flaky tests in this monorepo, with a bias for e2e (*.e2e-spec.ts). Use when tests fail after a change, when a suite passes alone but fails in the full run, on fixture/bootstrap drift, barrel-registration collisions, teardown/open-handle leaks, or runner/config drift. Triggers on "fix the tests", "e2e failing", "flaky test", "tests pass alone but not together".---# E2E FixerE2E is the default test tier here (`*.e2e-spec.ts`, real Nest app + supertest + SQLite). Most "failures"are environment/isolation, not logic — separate the two before editing source.## First: is it mine, pre-existing, or flaky?1. Run the failing suite **in isolation**: `corepack yarn vitest run --project e2e-packages "<suite-path>"`. Passes alone but fails in the full run → see the known rotating-404 flake in CHANGELOG.md before digging.2. Establish a baseline with `git stash` (keep node_modules) and re-run the suite. Same failure on clean HEAD → pre-existing, not your regression. Say so explicitly; do not "fix" pre-existing breakage silently.## Known failure classes and fixes- **Rotating full-run failures = host memory pressure, not a code bug.** Symptoms vary (unexpected 404, an expected 401 arriving as 404, `Parse Error: Expected HTTP/`) with a different victim suite each run. Tell: the failing run takes ~20x longer than a passing one (216s vs 10.8s) and the real failures are 30s timeouts with no HTTP response. Check free RAM / pageouts before touching any test. Diagnosis and evidence in CHANGELOG.md "Intermittent e2e failures". Fix the host (`--maxWorkers=2`, free memory) — never retry-loop, skip, or weaken assertions.- **Barrel registration collisions.** Importing a `domains/*/index` barrel registers `@CommandHandler`/ `@QueryHandler` in global Reflect metadata and breaks later Nest apps in the same process. Never import a barrel in an e2e file that boots an app. (The barrel-last sequencer is gone; no barrel-only specs exist.)- **Teardown / open handles.** A hanging suite process + `Parse Error: Expected HTTP/` from supertest = an app wasn't closed. Ensure `await app.close()` in `afterEach`/`afterAll`. `forceExit` is off on purpose — a leak hangs and gets reported instead of being masked.- **Fixture drift (v7→v8).** Symptoms: 500s, `Class extends value undefined`, `x is not a function`. Fix the fixture to the v8 pattern — `RepositoryModule.forFeature` for every entity, no duplicate `TypeOrmModule.forRoot`, correct `extras` (user/otp/role/federated/invitation). Reuse `packages/rockets-server-auth/src/__fixtures__/` and `__e2e__/helpers/`.- **Module-resolution after a bump.** `Cannot find module '@concepta/.../subpath'` → see **upstream-migrator**. Vitest resolves the package `exports` maps natively; if a subpath fails, the exports map itself is wrong.## Rules- New tests must be `*.e2e-spec.ts` unless a unit test is specifically justified. Coverage target ≥ 80% via `yarn test:e2e:cov`.- Fix the test setup or the real bug — never weaken an assertion or add `any` to make red go green.- Report the final tally honestly (passed/failed/skipped) and label each remaining failure as fixed, pre-existing, or flaky.

Frequently asked questions