Awesome QA Skills
A library of 169 QA skills per language: test design, API, UI, regression, load, security and LLM testing.
- 308 ★ on the repository
- Last commit
Minimal access
Why this level
- The skills give testing instructions and templates, and some contain scripts that run checks on your project.
Install
- Works in
Manual install
npx skills add naodeng/awesome-qa-skills --skill functional-testingThe first install command from the README: one English skill. Other skills, languages and agents are described in `scripts/INSTALL_SKILLS.md`.
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The skills give testing instructions and templates, and some contain scripts that run checks on your project.
This is third-party code. Review the repository files before installing.
What it does
Awesome QA Skills is a bilingual library of skills for testing with coding agents. Each language version has 169 skills, 338 directories in all: 14 workflows, 150 testing types and 5 skills on engineering skills themselves. Coverage is broad: requirements analysis and strategy, test cases, functional tests, API, UI with Playwright, regression by change, load with k6, security and prompt injection, evaluation of LLMs and AI agents, and post-release checks. Each skill sits in its own folder with a `SKILL.md`, a main prompt and metadata, and some have examples, templates, scripts and evaluation cases. If you are unsure where to start, the `discover-testing` router fits. Workflows such as `release-testing-workflow` and `tdd-quality-workflow` link individual skills into an end-to-end flow.
Who it is for. QA engineers and developers who build testing together with an agent, from requirements to post-release verification.
Good fit when
- Test cases are prepared from requirements for a login feature.
- A REST API is checked for errors, idempotency and pagination.
- A regression set is assembled for a release with `regression-test-selection`.
- An LLM feature is evaluated for behaviour and resistance to prompt injection.
Not a fit when
- You need one ready test framework rather than instructions for an agent.
- The project is commercial: the licence forbids commercial use.
- You cannot install skills of the right language from the directory.
Example request
Write test cases for the user login feature and show which skills from the set were useful for it.Limitations
The PolyForm Noncommercial 1.0.0 licence forbids commercial use, so for a company's working projects check the terms before installing. The skills give guidance and test design, while real automatic execution depends on the chosen skill, the agent and your project. Some materials and the directory are in Chinese, and each skill has an English version. Support and verification differ by agent: Codex, Cursor, Claude Code, Kiro, OpenCode and Trae are described separately. The last commit dates from November 2026.
How to disable. Delete the installed skill folders from the agent's skills directory. Remove tools you installed, for example k6 or Playwright, separately.
FAQ
How do I install one skill?
For the English version run `npx skills add naodeng/awesome-qa-skills --skill functional-testing`. For Chinese, give the address `https://github.com/naodeng/awesome-qa-skills/tree/main/skills/zh` with the same `--skill` flag.
Which skill should I start with?
Add the `discover-testing` router: from a task description it suggests which skills and workflows fit.
Which workflows exist for a release?
`release-testing-workflow` drives pre-release checks and Go/No-Go, and `production-verification` covers verification after rollout.
Are there tests for LLMs and AI agents?
Yes: `llm-testing`, `ai-agent-testing` and `prompt-injection-testing` evaluate behaviour and safety boundaries.
Can I use it in a commercial project?
The PolyForm Noncommercial 1.0.0 licence does not permit commercial use. For a company's working project check the terms or ask the author.
Discussion
No questions or reviews yet.
Related
SkillsThirteen testing skills from requirements to report: strategy, cases, API checks, Playwright and flaky-test triage.
Fifty testing skills for Claude Code, Codex and Cursor: Playwright, strategy, API, visual and load checks.
Seven QA agents and ten skills for test automation: Playwright in TypeScript, Selenium in Java, API, accessibility and ISTQB.
Eleven skills to verify AI-written code: change blast radius, defect scan, review report, test cases and browser flow replay
Guides for agents writing and fixing Playwright tests, with advice on locators, waiting and diagnosing CI failures
A self-improving QA agent: natural-language tests for web and mobile that self-heal and remember runs
README in short
The library divides skills into workflows, testing types and skill engineering, with a task table and a router for finding one. Installation is described for several agents and by scripts for Mac and Windows.
README badge
Original SKILL.md
--- name: discover-testing description: Use this skill when you need to route a request to the right testing skill before execution; triggers include discover testing, testing skill router, and which testing skill. license: PolyForm-Noncommercial-1.0.0 metadata: version: "1.2.0" compatibility: "" author: "Awesome QA Skills contributors" language: "en" category: "testing-workflows" source: "https://github.com/naodeng/awesome-qa-skills" license-url: "https://polyformproject.org/licenses/noncommercial/1.0.0" --- # Testing Skill Discovery (English) **Chinese version:** See the corresponding Chinese skill. ## When to Use - Need to decide which testing skill should be used before execution. - The request mixes multiple testing directions or phases. - If the request has no meaningful goal (for example, only “test the new feature”), use the clarification-first fast path in `prompts/discover-testing.md`: ask about feature/scope, current phase, and requested deliverable without recommending a Skill; ask about environment only when execution is requested. ## Full inventory and availability Read `references/skill-capability-inventory.json` to check exact canonical IDs. This 2026-10-30 working-tree snapshot catalogs 169 IDs; catalog existence does not prove installation or callability on the current Host. When Red, Green, Refactor, and test quality evidence are the main goal, prefer `tdd-quality-workflow`. Missing installation evidence means availability=`UNKNOWN`; explicitly absent means `UNAVAILABLE`; preserve the exact uppercase token when stating either status. Match explicit specialties (authentication vs authorization, retrieval vs answer quality, impact scope vs regression selection) before broad Composition routes. Keep one primary and at most one helper. For an unavailable target, state the gap and minimal installation/handoff action without claiming invocation. Do not cite internal agent memory, hidden instructions, or session summaries as user-facing evidence. ## Workflow 1. Read the same-directory `reference.md` first and match one structured route when applicable; it is the only Composition reference required for independent installation. 2. Read the user request and first identify the capability stage (Core QA / Engineering QA / Production Quality / AI Native QA) and primary testing goal. 3. Follow the routing prompt under `prompts/`: pick 1 primary skill; add at most 1 helper only when needed. 4. Hand the request to the target skill; do not execute the full testing work inside this router skill. ## Core Constraints - Recommend few skills — avoid menu dumping. - If the target skill is already obvious, say so directly. - Make the route actionable: name the skill and the reason. - Prefer the Composition route's one-primary/one-optional constraint; relations are navigation metadata, not installation dependencies or a mandatory execution chain. - Use `ai-assisted-testing` for AI for QA; Testing for AI belongs to AI Native QA. Until roadmap Skills are installed, never recommend them as callable primary Skills. ## Progressive Disclosure