What is the difference between Supervised and Delegated mode?

Supervised mode pauses to negotiate with the user when tools are missing, while Delegated mode proceeds autonomously, documenting any tool gaps in the final report.

Can it test command-line interfaces (CLI)?

Yes, it uses bash and shell capabilities to verify CLI behavior, execute curl commands for API validation, and inspect file system or log changes.

How does the QA skill report its findings?

It generates and updates a structured 'qa-progress.json' file which tracks every test scenario, its status, evidence of the test, and notes on any failures found.

What tools does the QA skill use for web testing?

It specifically uses the /browser skill (powered by Playwright) for headless automation, navigation, form filling, and monitoring network or console errors.

Does this skill replace standard unit tests?

No, the QA skill is designed to run after unit tests and linting have already passed. It focuses on the user experience, visual correctness, and integration scenarios that formal tests often miss.

Manual QA Testing

Name: Manual QA Testing
Author: inkeep

byinkeep

•

安全与测试

Verifies feature implementation from a user's perspective by performing end-to-end testing across browsers, APIs, and CLI tools.

The QA skill acts as a dedicated quality assurance engineer within Claude Code, focusing on the end-to-end verification that formal automated test suites often miss. It evaluates visual correctness, UX flows, and integration reality using a multi-tool approach that includes browser automation via Playwright, shell scripts for API validation, and desktop automation. By prioritizing human-centric factors like usability judgment and failure recovery, this skill ensures features are not just logically sound but provide a high-quality user experience before they reach production.

主要功能

01Visual and responsive layout audits using Playwright-based browser automation

02Integration reality checks against live APIs, databases, and system logs

03Structured QA reporting with automated generation of qa-progress.json

04Smart context extraction from SPEC files, PR diffs, and git history

05End-to-end user journey verification across web, mobile, and CLI surfaces

063 GitHub stars

使用场景

01Verifying a new UI feature renders correctly across different viewport sizes and handles edge cases gracefully

02Performing a comprehensive smoke test on a feature branch before merging a pull request

03Validating complex multi-step workflows that span both web interfaces and backend API endpoints

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add inkeep/team-skills qa

For use in Claude.ai and ChatGPT

Download Skill