Skip to main content
skill CX Codex

browse-qa — Claude Code skill: QA runs and setup

browse-qa is a skill that turns feature requests into explicit QA scenarios, runs them through browse, and saves stable flows for regression checks.

no license 148 installs ★ 5 updated 2026-08-31

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add ulpi-io/skills --skill browse-qa -a claude-code
Open on GitHub

What browse-qa does

Give browse-qa a ticket, URL, app target, or plain-language test request. It clarifies the target and platform when needed, then lays out expected outcomes for normal use, errors, edge cases, and exploratory checks. Ambiguity gets resolved before testing begins.

The skill delegates browser or simulator actions to browse, while it manages the test plan, evidence, and report. It records outcomes with supporting details such as a screenshot path or console error. Stable scenarios can become separate, named flows for later regression runs; noisy or incomplete recordings are left out. The browse skill is needed to interact with the target.

The report identifies the target, scenario results, evidence, saved flow files, and blockers. This makes it possible to review what happened without replaying the whole session. browse-qa is for QA planning and execution, not static code review or low-level browser control.

What it helps with

  • turn a feature request into test scenarios
  • exercise a real app flow against expected outcomes
  • capture evidence for failed or important checks
  • save stable scenarios as regression flows
  • report results and testing blockers

Example prompts for the agent

  • QA this checkout flow, cover error cases, and save stable regression flows.
  • Validate LINEAR-123 against its acceptance criteria and report each scenario's result.
  • Turn my manual checks for this page into rerunnable QA scenarios.

How to install browse-qa

1Skills CLI (Vercel)

npx skills add ulpi-io/skills --skill browse-qa -a claude-code

2Manually: put the files in place

~/.claude/skills/browse-qa/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/browse-qa/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about browse-qa

How do I install browse-qa in Claude Code?

Run npx skills add ulpi-io/skills --skill browse-qa -a claude-code. The skill file is at ~/.claude/skills/browse-qa/SKILL.md.

What does browse-qa do?

It plans and runs QA scenarios from a request, delegates app interaction to browse, and prepares a report with evidence. Stable scenarios can be saved as regression flows.

Does browse-qa interact with the browser itself?

No. It handles the QA plan and results, and uses the browse skill for browser or simulator interaction.

Similar skills

All skills for Testing and QA →

From the same set49

The whole set →
update-skill-learnings update-skill-learnings captures one validated rule about skill design or quality in a central learnings file. Use it when a session reveals guidance worth keeping. CX Codex +1 155installs start start is a Claude Code skill that routes incoming tasks, selects domain skills, evaluates sub-agent delegation, and limits context exploration before code edits. CX Codex +1 152installs pr-retro The pr-retro skill measures a feature branch against its base and reports merge-readiness with recommendations. Run it before reviewing a PR. CX Codex +1 150installs ast-grep This skill uses the ast-grep CLI to find code by syntax. Use it when a text or regex search cannot express the structure you need. CX Codex +1 148installs secrets Drives the secrets CLI: add, rotate, list and toggle local service credentials while values stay out of commands and chat. CX Codex +1 148installs run-parallel-agents-feature-debug run-parallel-agents-feature-debug is a Claude Code skill that concurrently diagnoses and fixes three or more independent failures in isolated worktrees after checking for a shared cause. CX Codex +1 148installs find-agents find-agents is a Claude Code skill for searching, installing and updating agents across 43+ editors and CLIs, converting each to the target client's format. CX Codex +1 148installs claude-review Launches an isolated second-opinion reviewer in a read-only worktree to inspect your diff without touching local files. Useful before merging risky pull requests. CX Codex +1 148installs

Project sponsors

Companies that keep this analytics open