Test Lab

10 skills inclus

• 01

  Test-Driven Development

  Fait coder l'agent en rouge, vert, refactor avec les commandes de test du dépôt, prouve chaque bug par un test et choisit la bonne taille de test.+

  sf-tdd-workflow/SKILL.md, de addyosmani/agent-skills, licence MIT.

  1---

  2name: sf-tdd-workflow

  3description: "Drives code changes with the red-green-refactor loop, using the repository's own test commands,…

  4---

  5

  6# TDD Workflow

  7

  8Write a failing test before the code that makes it pass. For a bug, reproduce it with a test before touching…

  9

  10This skill is the pragmatic, project-aware version of TDD: it starts by learning how this repository runs tes…

  11

  12## When to use

  13

  14- New logic or behavior

  15- Any bug fix (Prove-It pattern below)

  16- Changes to existing behavior, edge-case handling, anything that could break callers
• 02

  Strict Test-Driven Development

  Impose à l'agent le test d'abord sans exception : test lancé et vu échouer pour la bonne raison, code minimal, code écrit avant son test supprimé.+

  sf-strict-tdd/SKILL.md, de obra/superpowers, licence MIT.

  1---

  2name: sf-strict-tdd

  3description: "Enforces strict test-first discipline: no production code without a test that was run and seen…

  4---

  5

  6# Strict TDD

  7

  8Write the test first. Watch it fail. Write the minimal code to pass. Watch it pass.

  9

  10Core principle: if you did not watch the test fail, you do not know whether it tests the right thing. Followi…

  11

  12This skill is the uncompromising version. For a lighter, project-aware loop focused on picking test sizes and…

  13

  14## When to use

  15

  16Always, for:
• 03

  JavaScript Testing Patterns

  Fait écrire par l'agent des tests JS et TS avec Vitest ou Jest et Testing Library : unitaires, mocks légers, fabriques de données, intégration.+

  sf-js-testing/SKILL.md, de wshobson/agents, licence MIT.

  1---

  2name: sf-js-testing

  3description: "Writes and sets up JavaScript and TypeScript tests with Vitest or Jest and Testing Library: uni…

  4---

  5

  6# JavaScript and TypeScript Testing Patterns

  7

  8Concrete, copyable patterns for testing JS/TS code with the runner the project already uses. For the test-fir…

  9

  10## When to use

  11

  12- Writing unit or integration tests for JS/TS functions, classes, services, API routes

  13- Setting up or fixing `vitest.config.ts` / `jest.config.ts`, setup files, coverage

  14- Mocking modules, `fetch`, timers, or injected dependencies

  15- Testing React components and hooks with Testing Library

  16
• 04

  Web App Testing

  Fait tester par l'agent une app web locale avec de courts scripts Playwright Python, serveurs de dev lancés et arrêtés par un script fourni.+

  sf-webapp-testing/SKILL.md, de anthropics/skills, licence Apache 2.0.

  1---

  2name: sf-webapp-testing

  3description: "Tests and inspects local web applications by writing short Python Playwright scripts, with a he…

  4---

  5

  6# Web App Testing with Python Playwright

  7

  8Verify a local web app by driving a real headless browser from a Python script. This is for quick, scripted c…

  9

  10## Requirements

  11

  12- Python 3 with the `playwright` package and a Chromium build (`pip install playwright` then `playwright inst…

  13- The app's own start command (read `package.json` scripts or the README).

  14

  15## Bundled files

  16
• 05

  Playwright CLI

  Fait piloter un vrai navigateur à l'agent avec playwright-cli : snapshots d'accessibilité, actions par ref, réseau, traces, code de test généré.+

  sf-playwright-cli/SKILL.md, de microsoft/playwright-cli, licence Apache 2.0.

  1---

  2name: sf-playwright-cli

  3description: "Drives a real browser from the shell with Microsoft's playwright-cli: open pages, read accessib…

  4allowed-tools: Bash(playwright-cli:*) Bash(npx playwright:*) Bash(npx --no-install playwright:*)

  5---

  6

  7# Browser Automation with playwright-cli

  8

  9`playwright-cli` keeps a browser session open between shell commands. Each command returns the page status an…

  10

  11Related skills: `sf-playwright-e2e` for designing and maintaining the test suite itself; `sf-webapp-testing`…

  12

  13## Setup

  14

  151. Check for a local Playwright first: `npx --no-install playwright --version`. If it works, use `npx playwri…

  162. Otherwise check for a global `playwright-cli`. If neither exists, ask the user before installing: `npm ins…
• 06

  Playwright E2E Best Practices

  Fait écrire et stabiliser par l'agent des suites Playwright Test : locators robustes, assertions web-first, fixtures, auth, mocks, tests instables, CI.+

  sf-playwright-e2e/SKILL.md, de currents-dev/playwright-best-practices-skill, licence MIT.

  1---

  2name: sf-playwright-e2e

  3description: "Writes, structures, debugs and scales Playwright Test suites (@playwright/test): resilient loca…

  4---

  5

  6# Playwright Test Suites

  7

  8Build Playwright tests that fail only when the product is broken. This file holds the rules that apply to eve…

  9

  10Related skills: `sf-playwright-cli` to drive a browser interactively or heal a test against the live page; `s…

  11

  12## When to use

  13

  14- Writing new E2E, component, API or visual tests with `@playwright/test`

  15- Fixing failing, flaky or slow Playwright tests

  16- Structuring a suite: fixtures, page objects, test data, tags, projects
• 07

  Browser Testing with DevTools

  Fait vérifier par l'agent le comportement réel dans le navigateur via le serveur MCP Chrome DevTools : DOM, console, réseau, traces, accessibilité.+

  sf-browser-devtools/SKILL.md, de addyosmani/agent-skills, licence MIT.

  1---

  2name: sf-browser-devtools

  3description: "Verifies and debugs browser behavior with real runtime data through the Chrome DevTools MCP ser…

  4---

  5

  6# Browser Verification with Chrome DevTools MCP

  7

  8Give the agent eyes on the running page. Instead of reasoning about what the code should render, look at what…

  9

  10Related skills: `sf-tdd-workflow` for the automated test that should guard the fix; `sf-playwright-e2e` for a…

  11

  12## When to use

  13

  14- Any change that renders in a browser (component, page, CSS, client logic)

  15- UI bugs: layout, styling, interaction, missing elements

  16- Console errors or warnings, failed or wrong network calls, CORS
• 08

  Agent Browser

  Fait piloter un vrai navigateur à l'agent avec agent-browser : snapshots à refs @eN, formulaires, attentes fiables, extraction, captures, QA exploratoire.+

  sf-agent-browser/SKILL.md, de vercel-labs/agent-browser, licence Apache 2.0.

  1---

  2name: sf-agent-browser

  3description: Automates a real Chrome browser from the shell with the agent-browser CLI, using accessibility-t…

  4allowed-tools: Bash(agent-browser:*), Bash(npx agent-browser:*)

  5---

  6

  7# Agent Browser

  8

  9Drive Chrome or Chromium over CDP with the `agent-browser` CLI. Snapshots of the accessibility tree give comp…

  10

  11## When to use

  12

  13- Interacting with a live site or local app: navigate, click, fill, extract, screenshot.

  14- Exploratory testing and bug hunts with reproduction evidence (`references/dogfood.md`).

  15

  16For Playwright test suites, use `sf-playwright-e2e` or `sf-playwright-cli`; for Chrome DevTools debugging, `s…
• 09

  Debugging and Error Recovery

  Fait arrêter la chaîne à l'agent dès qu'une erreur survient : reproduire, localiser, réduire, corriger la cause, ajouter un test de non-régression, vérifier.+

  sf-error-recovery/SKILL.md, de addyosmani/agent-skills, licence MIT.

  1---

  2name: sf-error-recovery

  3description: Stops the line when something breaks and runs a structured triage (reproduce, localize, reduce,…

  4---

  5

  6# Debugging and Error Recovery

  7

  8When something breaks, stop adding features, preserve the evidence, and find the root cause with a fixed sequ…

  9

  10## When to use

  11

  12Any unexpected failure during development or after a change. For long, hard investigations with hypotheses an…

  13

  14## Stop the line

  15

  16```
• 10

  Code Review and Quality

  Fait relire un changement par l'agent sur 5 axes (exactitude, lisibilité, architecture, sécurité, perf), tests vérifiés, constats classés et verdict.+

  sf-quality-review/SKILL.md, de addyosmani/agent-skills, licence MIT.

  1---

  2name: sf-quality-review

  3description: Reviews code changes across five axes (correctness, readability and simplicity, architecture, se…

  4---

  5

  6# Code Review and Quality

  7

  8Every change gets reviewed before merge. Approve a change when it clearly improves overall code health, even…

  9

  10## When to use

  11

  12- Before merging any change, including your own.

  13- After a feature or a bug fix (review the fix and its regression test).

  14- When another agent or model produced code you must evaluate.

  15

  16For a lighter review focused on the diff's intent, `sf-change-review` is an alternative; for handling feedbac…

Extraits. Les dossiers complets sont livrés après le paiement.

Installation