Exécutez n'importe quel Skill dans Manus
en un clic

Exécutez n'importe quel Skill dans Manus en un clic

write-tests

Étoiles0

Forks0

Mis à jour9 juin 2026 à 03:20

Write and run tests that catch real regressions, match the project's existing harness and style, and cover edge/error paths. Use when adding tests, improving coverage, or after implementing a feature.

Installation

Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.

Exécuter dans Manus

Source

linder3hs

linder3hs/setup-ai

Ouvrir le dépôt GitHub Voir les dépôts du créateur

Téléchargement

Exécuter dans Manus

Métiers associésSOC

Basé sur la classification professionnelle SOC

Analystes en assurance qualité des logiciels et testeursProfessions informatiques et mathématiques·SOC 15-1253

SKILL.md

readonly

name	write-tests
description	Write and run tests that catch real regressions, match the project's existing harness and style, and cover edge/error paths. Use when adding tests, improving coverage, or after implementing a feature.

Write tests

Tests that catch regressions, not coverage padding.

Steps

Detect the harness. Read package.json scripts + devDeps to find the runner and command (Vitest, Jest, Playwright, node:test). Match the existing test file location and naming. Never introduce a new framework.
Pick targets. Business logic, edge cases, error paths, regressions for known bugs. Skip trivial getters and framework glue.
Write. One behavior per test. Descriptive name. Arrange-act-assert. Cover happy path + the boundaries (empty, null, max, error). Mock only true external boundaries (network, clock, fs).
Run. Execute the suite (npm test / pnpm test / the detected command). Report output verbatim. A test not run is not done.

Rule

If a test fails because of a real product bug, report the bug — never weaken the assertion to make it pass.

Output

Files added (paths), command run, pass/fail result. State honestly what remains untested.

Plus depuis ce dépôt

même dépôt

plan-feature

linder3hs/setup-ai

Plan a feature end-to-end before coding — clarify intent, explore existing code for reuse, design the approach, list files to change, and define how to verify. Use before implementing any non-trivial feature or when asked to plan or design work.

2026-06-090

qa-browser

linder3hs/setup-ai

QA a web app/UI in a real browser — navigate flows, verify state, catch console/network errors, take screenshots as evidence. Use when asked to QA, test the site, verify a deployment, or confirm a UI feature works end-to-end.

2026-06-090

review-diff

linder3hs/setup-ai

Review the current diff/PR for correctness bugs, reuse, simplification, and efficiency. Use before merging, after finishing a change, or when asked to review code.

2026-06-090

scaffold-project

linder3hs/setup-ai

Bootstrap a new project, module, component, or route following existing conventions (or sensible Next.js/React/TS defaults for a fresh repo). Use when asked to scaffold, set up, or create a new component/route/module/project.

2026-06-090

security-audit

linder3hs/setup-ai

Audit code for exploitable security vulnerabilities — injection, auth/authorization flaws, secrets, trust-boundary violations, OWASP Top 10. Use when reviewing security-sensitive code (auth, payments, input handling) or when asked for a security audit. Defensive use only.

2026-06-090

name	write-tests
description	Write and run tests that catch real regressions, match the project's existing harness and style, and cover edge/error paths. Use when adding tests, improving coverage, or after implementing a feature.

Write tests

Tests that catch regressions, not coverage padding.

Steps

Detect the harness. Read package.json scripts + devDeps to find the runner and command (Vitest, Jest, Playwright, node:test). Match the existing test file location and naming. Never introduce a new framework.
Pick targets. Business logic, edge cases, error paths, regressions for known bugs. Skip trivial getters and framework glue.
Write. One behavior per test. Descriptive name. Arrange-act-assert. Cover happy path + the boundaries (empty, null, max, error). Mock only true external boundaries (network, clock, fs).
Run. Execute the suite (npm test / pnpm test / the detected command). Report output verbatim. A test not run is not done.

Rule

If a test fails because of a real product bug, report the bug — never weaken the assertion to make it pass.

Output

Files added (paths), command run, pass/fail result. State honestly what remains untested.