Top 9 Open-Source Claude Skills for Testing & QA in 2026

Agent skills have quietly become one of the fastest-growing categories in software testing. Since Anthropic launched Agent Skills as an open standard in October 2025, developers have been packaging testing methodologies, QA workflows, and framework expertise into simple SKILL.md files that Claude Code, Codex, Cursor, and Gemini CLI can load on demand. In 2026, the ecosystem includes dedicated QA skill libraries with dozens of skills, TDD methodology packs with hundreds of thousands of installs, and framework-specific skills from vendors like Microsoft, Currents, Vercel, and Trail of Bits. The problem? Quality varies wildly. Published research suggests roughly 26% of agent skills carry at least one vulnerability, and many skills claim capabilities their bundled code can't deliver. That's why we built this list: to highlight open-source Claude skills for testing and QA that are genuinely worth installing in 2026, along with a free way to evaluate them safely. Whether you're a developer wiring Playwright into your CI pipeline or a QA lead exploring AI-assisted test generation, here are the nine picks that stand out this year.

#1 SkillzBench — Editors' choice

SkillzBench is AIQ Labs' free, independent directory of AI agent skills plus a free MCP tool that connects any agent to the skills you choose — and it takes the top spot because it solves the exact problem this list creates. With thousands of skills floating around GitHub and community marketplaces, nobody can tell which ones actually help and which are risky. SkillzBench is being built to change that: it continuously crawls and dedupes skills from GitHub, pins every version to a commit SHA, validates them, checks licenses, runs security scans, and applies quality judging before anything appears in the directory. What truly sets SkillzBench apart is its benchmarking approach. Every skill gets tested in a sandbox with the same model, both with and without the skill, and the resulting Lift — measured in percentage points — is published separately from ratings. A trust score from 0 to 100 combines security, quality, license, maintenance, and community signals, and money never touches scores. The platform also tracks upstream changes, re-scans, re-benchmarks, and notifies you when a skill you rely on changes. Everything on SkillzBench is free: the directory, the MCP connector, benchmarks, and alerts. The platform is currently in active development with a waitlist live, so you can't install skills from it today — but for anyone serious about using testing and QA skills safely in 2026, joining the waitlist is the smartest first move.

Best for: Developers, QA engineers, and agent builders who want to discover, verify, and safely install testing and QA skills for Claude Code, Codex, Cursor, or Gemini CLI without guessing which ones are trustworthy

Pricing: Free (waitlist currently open; the directory, MCP connector, benchmarks, and alerts are all free)

Key features

• Free, independent directory of AI agent skills crawled and deduped from GitHub
• Security scanning, license checking, and validation for every skill
• Every skill version pinned to a commit SHA for reproducibility
• Sandbox benchmarking that publishes Lift in percentage points, with and without the skill
• Trust score (0-100) built from security, quality, license, maintenance, and community signals
• Free MCP server plus per-agent loadouts — agents only see the skills you toggle on
• Upstream change tracking with re-scans, re-benchmarks, and notifications
• Everything free, funded by advertising that never influences scores or listings

Pros

• Completely free — no paid tiers for the directory, MCP, or benchmarks
• Addresses the ~26% skill vulnerability problem with security scans and license checks
• Lift benchmarking shows actual performance impact, not marketing claims
• Works across multiple agents via the free MCP connector, not just Claude Code

Cons

• Still pre-launch — the product is not yet publicly available
• Waitlist required to get early access
• Benchmark coverage will take time to build across the full skill ecosystem

Visit SkillzBench

#2 qa-skills

qa-skills is a library of 42 QA and test-automation skills built for AI coding agents, shared on GitHub in May 2026 and licensed under MIT. It is arguably the most comprehensive dedicated QA skill pack in the ecosystem, organized across ten categories: Playwright and Cypress automation (end-to-end testing, page object models, CI integration), API, unit, and mobile testing (REST/GraphQL, Jest/Vitest, Appium/Detox), and infrastructure skills covering CI/CD pipelines, test environments, and contract testing. Where qa-skills really differentiates itself is in its AI-augmented QA category — skills like ai-test-generation, ai-bug-triage, test-reliability, and ai-qa-review that teach agents how AI can help with the testing process itself, not just run tests. It also covers specialized domains including accessibility (WCAG 2.1), security (OWASP), visual regression, and performance, plus production-oriented skills for synthetic monitoring, trace-based testing, and chaos engineering. All 42 skills follow the Agent Skills Standard format, so they work drop-in with Claude Code, OpenAI Codex, Cursor, Gemini CLI, and any compatible runtime. According to the author, the pack was built from patterns across six-plus production QA setups. Installation is a one-liner via npx. For QA teams that want a single, broad-coverage skill library rather than piecing together individual skills, qa-skills is the most complete open-source starting point in 2026.

Best for: QA teams and test engineers who want a comprehensive, drop-in skill library covering the full testing lifecycle for any Agent Skills Standard-compatible agent

Pricing: Free (MIT license)

Key features

• 42 QA and test-automation skills across 10 categories
• Playwright and Cypress skills for E2E testing, POM, and CI integration
• API, unit, and mobile testing coverage (REST/GraphQL, Jest/Vitest, Appium/Detox)
• AI-augmented QA skills: ai-test-generation, ai-bug-triage, test-reliability, ai-qa-review
• Specialized skills for accessibility (WCAG 2.1), security (OWASP), visual regression, and performance
• Production and observability skills: synthetic monitoring, trace-based testing, chaos engineering
• Follows the Agent Skills Standard — works with Claude Code, Codex, Cursor, and Gemini CLI
• MIT-licensed and installable via npx skills add

Pros

• Broadest dedicated QA skill coverage available — 42 skills in one pack
• Unique AI-augmented QA category covering test generation and bug triage
• Cross-agent compatibility via the Agent Skills Standard
• Built from real production QA setups, per the author

Cons

• Breadth means you'll likely use only a subset — extra skills add token overhead
• Individual skills vary in depth compared to single-purpose alternatives
• No built-in security vetting or benchmarking of the skills themselves

Visit qa-skills

#3 test-driven-development (obra/superpowers)

The test-driven-development skill from the obra/superpowers repository is the single highest-installed testing skill in the Claude skills ecosystem, with roughly 167,000 installs and a 4/5 quality rating according to 2026 skill-market data. It doesn't run tests and it doesn't wrap a framework — instead, it encodes the TDD discipline itself: write the test first, watch it fail, then write the code that makes it pass. The skill is intentionally short, because the discipline is simple, but the effect is profound: instead of letting Claude skip ahead to the implementation it 'knows' is right, the skill forces the agent to produce a failing test before any production code exists. It's part of the broader Superpowers framework, which has grown to hundreds of thousands of GitHub stars and enforces a structured multi-phase development workflow with mandatory TDD at every stage. For testing and QA purposes, this skill changes agent behavior in a way that framework wrappers don't — it enforces a methodology rather than executing commands. That's why it dominates the testing category by install volume: developers using Claude Code for testing want the discipline baked in, not just another test runner. Install it before picking any Playwright wrapper, and pair it with a framework-specific skill if you need execution capabilities too.

Best for: Developers and teams who want Claude Code to follow strict test-driven development on every feature and bugfix

Pricing: Free (open source)

Key features

• Enforces test-first development: failing test before implementation code
• Highest-installed testing skill in the ecosystem (~167K installs, 4/5 quality)
• Part of the widely adopted Superpowers skills framework
• Short, focused SKILL.md that loads only when relevant
• Changes agent methodology rather than wrapping a test runner
• Pairs naturally with framework-specific skills like Playwright tools

Pros

• Proven popularity signal with strong install volume and quality rating
• Genuinely changes agent behavior — tests actually get written first
• Lightweight and focused on one job
• Backed by the actively maintained Superpowers ecosystem

Cons

• Doesn't execute tests — you still need a runner or framework skill
• Strict TDD can feel heavy for quick prototypes or throwaway scripts
• No built-in security vetting

Visit test-driven-development (obra/superpowers)

#4 Playwright Best Practices (Currents)

Published by Currents, a Playwright observability vendor that runs thousands of test suites at scale, this skill packages hard-won Playwright expertise into a single SKILL.md. With roughly 33,000 installs and a 3/5 quality rating per 2026 skill-market data, it stands out from generic Playwright CLI wrappers because of its anti-flake focus — flaky tests being the single biggest reason teams abandon end-to-end testing. According to its listing, the skill covers reliable selector strategies, parallel execution configuration, CI setup, and flake prevention patterns. Because Currents observes real-world Playwright failures across many teams, the guidance reflects actual failure modes rather than documentation rehashes — the vendor knows exactly where Playwright suites break in production CI environments. For teams building or maintaining Playwright suites with Claude Code, this skill functions as an embedded senior reviewer: when the agent writes or modifies tests, it follows the practices that keep suites stable. It pairs especially well with the official Microsoft Playwright CLI skill (which executes tests) and a TDD methodology skill (which ensures tests get written in the first place). If your E2E suite has ever gone red for no apparent reason, this skill addresses the root causes.

Best for: Teams using Playwright for end-to-end testing who want Claude Code to write stable, CI-friendly tests that don't flake

Pricing: Free (open source)

Key features

• Playwright best practices from a vendor running thousands of suites at scale
• Reliable selector guidance to reduce brittle tests
• Parallel execution and CI configuration patterns
• Flake prevention — the key differentiator over generic CLI wrappers
• ~33K installs with a 3/5 quality rating
• Works as an embedded reviewer when Claude writes or edits Playwright tests

Pros

• Vendor-backed expertise from real-world Playwright observability at scale
• Directly targets the flakiness problem that kills E2E adoption
• Practical CI and parallel-execution guidance
• Complements the official Playwright CLI skill well

Cons

• Playwright-only — no coverage for Cypress or other frameworks
• Doesn't execute tests itself
• Quality rating (3/5) suggests room for improvement in depth

Visit Playwright Best Practices (Currents)

#5 Playwright CLI (Microsoft)

This is the official Playwright CLI skill published by Microsoft, the company that maintains Playwright itself. With roughly 27,000 installs and a 3/5 quality rating according to 2026 skill-market data, its main advantage is authority: because it comes from the Playwright team, the API mapping is authoritative — when Claude generates test scripts, runs assertions, or manages browser contexts, it's working from accurate, first-party knowledge of the framework. According to its listing, the skill enables Claude to generate Playwright test scripts, execute assertions, and handle browser contexts directly from the agent session. That makes it the execution layer of a Playwright testing setup: pair it with the Currents Best Practices skill to ensure the tests it generates aren't flaky, and with a TDD skill to ensure tests are written before implementation code. Microsoft's backing also means the skill tracks the framework as it evolves — a meaningful advantage over community wrappers that can drift out of date as Playwright releases new versions. For any team standardizing on Playwright in 2026, this skill is the natural default for letting Claude Code actually run and interact with your test suite rather than just advise about it.

Best for: Teams on Playwright who want Claude Code to generate and execute tests with first-party, maintainer-accurate knowledge of the API

Pricing: Free (open source)

Key features

• Official Playwright skill published by Microsoft
• Generates Playwright test scripts from agent sessions
• Runs assertions and manages browser contexts
• Authoritative API mapping from the framework's maintainer
• ~27K installs with a 3/5 quality rating
• Tracks the framework as Microsoft updates Playwright

Pros

• Maintainer-published, so API knowledge is authoritative and current
• Covers the full execution loop: scripts, assertions, browser contexts
• Backed by Microsoft's ongoing Playwright development
• Pairs cleanly with best-practices and TDD skills

Cons

• Playwright-only — not useful for Selenium, Cypress, or mobile stacks
• Doesn't include anti-flake methodology guidance
• Quality rating (3/5) indicates average depth

Visit Playwright CLI (Microsoft)

#6 Trail of Bits Skills

Trail of Bits, one of the most respected names in security auditing, publishes a set of security-focused skills for Claude Code that bring professional-grade static analysis into the testing workflow. According to 2026 coverage of the skills ecosystem, the pack includes CodeQL and Semgrep static analysis capabilities, differential review, and supply-chain audit skills — added to major skill roundups in September 2026 based on the official Trail of Bits GitHub repository. For testing and QA purposes, these skills fill a gap most testing skills ignore: security testing. While frameworks like Playwright verify that features work, Trail of Bits skills help agents hunt for vulnerabilities — querying code with CodeQL, running Semgrep rules for known bug patterns, and auditing dependencies for supply-chain risks. This aligns directly with OWASP-style security testing categories that mature QA programs require. Because it comes from Trail of Bits — a firm whose reputation depends on security rigor — it's also one of the safer third-party skill sources to install, a point explicitly noted in 2026 skills guides that flag undocumented network calls in lesser-known community skills as a real risk. For teams building a comprehensive agent-assisted QA stack, this is the security layer.

Best for: Security-conscious engineering teams adding security testing and vulnerability detection to their agent-assisted QA workflow

Pricing: Free (open source)

Key features

• Security audit skills from renowned security firm Trail of Bits
• CodeQL static analysis for vulnerability detection
• Semgrep rule-based code scanning
• Differential review capabilities for comparing code changes
• Supply-chain audit skills for dependency risk
• Considered a trusted source alongside Anthropic's official and Vercel skills

Pros

• Backed by one of the most trusted names in security research
• Covers vulnerability detection, not just functional testing
• Includes supply-chain auditing — increasingly critical in 2026
• Explicitly recommended as a safe default skill source

Cons

• Security-focused only — no functional, UI, or E2E testing coverage
• Requires familiarity with CodeQL/Semgrep concepts to interpret results
• Static analysis may surface false positives that need human triage

Visit Trail of Bits Skills

#7 E2E Testing Patterns

The E2E Testing Patterns skill from the wpank repository covers reliable end-to-end testing with both Playwright and Cypress — which is exactly what makes it useful. Most testing skills in the 2026 ecosystem are Playwright-only; this one explicitly supports teams whose stacks include Cypress, covering critical user journey coverage, flake elimination, and CI/CD integration patterns, according to its skill-market listing. With roughly 6,500 installs, it's smaller than the big Playwright skills, but its dual-framework coverage fills a real gap: many organizations run both Playwright and Cypress (or are migrating between them), and a skill that understands both frameworks' idioms keeps the agent consistent across the codebase. The flake-elimination focus mirrors the industry's biggest E2E pain point — tests that pass locally and fail in CI for timing and selector reasons. The skill also addresses critical user journey coverage, helping agents prioritize the flows that actually matter to users rather than generating exhaustive but low-value test coverage. For Cypress-based teams, 2026 skill guides recommend this over the Playwright-only alternatives. It's a focused, practical skill that does one job — reliable E2E patterns — without trying to be a full QA methodology.

Best for: Teams running Cypress (or a mix of Cypress and Playwright) who want reliable E2E testing patterns applied by their coding agent

Pricing: Free (open source)

Key features

• Covers both Playwright and Cypress — rare dual-framework support
• Critical user journey coverage patterns
• Flake elimination guidance for stable CI runs
• CI/CD integration patterns for E2E suites
• ~6.5K installs in the 2026 ecosystem
• Focused single-purpose skill with a lean footprint

Pros

• One of the few skills covering Cypress alongside Playwright
• Targets flake elimination — the top E2E pain point
• Prioritizes user-critical journeys over exhaustive low-value coverage
• Recommended for Cypress stacks by 2026 skill guides

Cons

• Smaller community and install base than the major Playwright skills
• No quality rating published
• Doesn't execute tests — pattern guidance only

Visit E2E Testing Patterns

#8 Vercel Agent Skills

Vercel Engineering publishes a set of agent skills that, while broader than pure testing, include capabilities directly relevant to QA: a UI audit skill that checks interfaces against more than 100 accessibility and UX rules, and over 40 React and Next.js performance rules. According to 2026 skills coverage, these were added to major skill roundups in September 2026 from Vercel's official GitHub repositories. For frontend QA in particular, the accessibility auditing capability is significant — accessibility testing is a specialized QA discipline (aligned with WCAG standards) that most testing skills skip entirely. The UI audit skill lets Claude Code review frontend code against a substantial ruleset covering accessibility and UX issues before they reach users, functioning as an automated first-pass review. The performance rules complement this by catching React and Next.js anti-patterns that degrade real-world user experience. Vercel's involvement matters for trust and maintenance: as the company behind Next.js, its knowledge of frontend failure modes is first-party, and the skills are explicitly cited as safe defaults alongside Anthropic's official skills and Trail of Bits in 2026 guides. For teams shipping React and Next.js applications, these skills add a frontend-quality QA layer that pure test-runner skills can't provide.

Best for: React and Next.js teams who want automated accessibility, UX, and performance auditing built into their agent-assisted QA process

Pricing: Free (open source)

Key features

• UI audit skill checking 100+ accessibility and UX rules
• 40+ React and Next.js performance rules from Vercel Engineering
• First-party frontend expertise from the Next.js maintainers
• Cited as a safe, trusted skill source in 2026 guides
• Automated pre-release frontend quality review
• Complements functional test-runner skills with quality-rule enforcement

Pros

• Covers accessibility testing — a gap in most testing skill packs
• Maintainer-grade frontend knowledge from Vercel
• Trusted-source status reduces supply-chain risk
• Practical rule-based reviews rather than vague advice

Cons

• Frontend-focused — no API, mobile, or backend testing coverage
• Rule auditing isn't a substitute for runtime E2E testing
• Primarily valuable if you're on React/Next.js

Visit Vercel Agent Skills

#9 CI/CD Pipeline (gitgoodordietrying)

Tests don't help if they don't run on every pull request — and that's the gap the CI/CD Pipeline skill fills. According to its 2026 skill-market listing, the skill writes GitHub Actions YAML that wires your test runner into continuous integration, covering testing, deployment, and release workflows. With roughly 4,200 installs, it's a smaller skill, but it addresses a dependency every other testing skill has: someone has to actually configure the pipeline. In practice, this skill is the connective tissue of an agent-assisted QA stack. You pick a test runner (Playwright, Cypress, Jest, pytest), adopt a TDD methodology skill to ensure tests get written, and then use this skill to make sure the whole thing executes automatically on every commit — generating the workflow configuration that runs tests, gates merges, and triggers releases. 2026 skill guides recommend using it after picking your test runner, which is the right sequencing: it doesn't choose frameworks or write tests, it operationalizes them. For solo developers and small teams without dedicated DevOps expertise, having Claude Code generate correct, working GitHub Actions pipelines removes one of the most common reasons testing initiatives stall — the tests exist but nothing runs them consistently. It's a narrow skill doing an essential job.

Best for: Solo developers and small teams who need their tests to actually run in CI but lack dedicated DevOps expertise

Pricing: Free (open source)

Key features

• Generates GitHub Actions YAML for testing workflows
• Covers testing, deployment, and release pipelines
• Wires any test runner into CI on every PR
• ~4.2K installs in the 2026 ecosystem
• Designed to be used after selecting a test runner
• Removes the DevOps configuration bottleneck for small teams

Pros

• Solves the 'tests exist but nothing runs them' problem
• GitHub Actions is the dominant CI platform for open-source and small teams
• Complements rather than duplicates test-runner and TDD skills
• Simple, focused scope

Cons

• GitHub Actions only — no GitLab CI, CircleCI, or Jenkins coverage
• No published quality rating
• Configuration generation may need manual review for complex pipelines

Visit CI/CD Pipeline (gitgoodordietrying)

Conclusion

Open-source Claude skills have matured from novelty to genuine QA infrastructure in 2026. The strongest setups combine layers: a methodology skill like test-driven-development to ensure tests get written first, framework skills like the Playwright CLI and Currents Best Practices to generate and run stable tests, specialized packs like qa-skills and Trail of Bits for security and specialized testing, and the CI/CD Pipeline skill to make it all run automatically. But before installing any of them, remember the ecosystem's dirty secret: research suggests roughly 26% of agent skills carry at least one vulnerability, and quality varies enormously. That's exactly why SkillzBench exists — a free, independent directory that security-scans, license-checks, and benchmarks every skill, publishing trust scores and Lift data so you can see what each skill actually delivers. The platform is in active development and everything will be free: the directory, the MCP connector, benchmarks, and alerts. Skills, proven. Join the SkillzBench waitlist today to be first in line when it launches, and start building a testing stack you can actually trust.

Frequently asked questions

What is a Claude skill for testing and QA?

A Claude skill is a folder containing a SKILL.md file — instructions plus optional scripts — that Claude Code loads on demand when a task matches the skill's description. Testing and QA skills package expertise like TDD methodology, Playwright best practices, security scanning, or CI/CD configuration so your coding agent applies them automatically. Because skills use progressive disclosure (only the name and description load at startup), you can install many without bloating your context window.

Are open-source Claude skills safe to install?

Not automatically. Published research found that roughly 26% of agent skills carry at least one vulnerability, and 2026 skills guides warn about undocumented network calls in bundled scripts from lesser-known sources. Skills from trusted publishers — Anthropic's official repository, Microsoft, Vercel, Trail of Bits, and Currents — are considered safer defaults. This is the core problem SkillzBench is being built to solve: it security-scans, license-checks, and pins every skill version to a commit SHA before listing it, with a 0-100 trust score combining security, quality, license, maintenance, and community signals.

Do these testing skills work with agents other than Claude Code?

Many do. Skills that follow the Agent Skills Standard — like the qa-skills library — work drop-in with Claude Code, OpenAI Codex, Cursor, Gemini CLI, and any compatible runtime. SkillzBench is also building a free MCP tool that connects any agent to the skills you choose, with per-agent loadouts so each agent only sees the skills and tools you've enabled for it.

What does SkillzBench cost?

Everything on SkillzBench is free: the skills directory, the MCP connector, benchmarks, and change alerts. The platform is funded by advertising shown in and around its blog articles, and ad money never influences scores or listings. The product is currently pre-launch with a live waitlist, so join now to get access when it becomes available.

Which testing skill should I install first?

Start with a methodology skill like test-driven-development from obra/superpowers — it's the highest-installed testing skill in the ecosystem and changes how your agent writes code, not just how it runs tests. Then add a framework-specific skill matching your stack (the Microsoft Playwright CLI skill for Playwright teams, E2E Testing Patterns for Cypress teams), and finish with the CI/CD Pipeline skill so your tests actually run on every pull request.

Can Claude skills replace my existing QA tools and frameworks?

No — and the honest 2026 consensus is that they shouldn't try. Skills like the Playwright CLI execute tests through frameworks you already have, and methodology skills enforce disciplines like TDD. AI handles the repetitive work: test generation, flake prevention, configuration, and static analysis. But understanding business risk, judging ambiguous results, and knowing which bugs matter to users still require human expertise. Think of skills as amplifying your QA engineers, not replacing them.

How do I know if a testing skill actually improves my results?

Look for evidence, not marketing. Most skills publish no performance data at all. SkillzBench is being built specifically to fix this: it benchmarks every skill in a sandbox with the same model, run both with and without the skill, and publishes the Lift in percentage points — always shown separately from ratings. Until then, run your own before-and-after comparison on a real project, and check community signals like install counts, quality ratings, and the publisher's track record.