agentleFS
Sign inStart free
matthullmatthull/my-skills/skills/workflow/manual-qa/SKILL.md

manual-qa

Skill0 starsChanged 55 days agoMITView on GitHub

Claude QA Engineer procedures — automated AND "manual" testing via tools before human UAT. Load this skill when Claude must verify features end-to-end via tools rather than deferring to human.

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 8 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Claude is the QA Engineer: All testing (automated AND "manual") happens before handoff to human.

    line 8
  2. Distinctive:

    If Claude cannot complete verification because tools are unavailable:

    line 20
  3. Distinctive:

    Test only happy path

    line 89
  4. Distinctive:

    Use available tools for visual verification

    line 92
  5. Distinctive:

    Test with production-like data volumes

    line 93
  6. Distinctive:

    Verify both UI (via screenshots) and database state

    line 94
  7. Distinctive:

    Test with multiple user roles/permissions

    line 95
  8. Distinctive:

    Document unexpected behavior found during testing

    line 96

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Manual QA Practice

Claude is the QA Engineer. All testing (automated AND "manual") happens before handoff to human.

"Manual" testing means using tools — not deferring to human:

  • Mobile: Waydroid + ADB screenshots, Maestro flows
  • Web: Browser automation, DevTools console, network inspection
  • API: curl/httpie, console verification, response inspection
  • Backend: Rails console, database queries, log inspection

CRITICAL: Tools Unavailable Protocol

If Claude cannot complete verification because tools are unavailable:

  1. IMMEDIATELY STOP — Do not proceed without verification
  2. Do NOT hand off with "please manually verify"
  3. Report clearly: What verification is needed, what tool would do it, why unavailable
  4. Ask for guidance: Setup help, alternative approach, defer task

You MUST NEVER:

  • Hand off with "please manually test [X]" when Claude should have tested it
  • Skip verification and hope for the best
  • Assume human will catch issues Claude should have caught

QA Gate Checklist

Claude completes ALL before handoff:

  • Automated tests passing (Loop 1 + Loop 2)
  • "Manual" verification complete via tools (Loop 3)
  • All edge cases tested
  • Error states verified (trigger errors, confirm behavior)
  • Data persistence verified (create, query, confirm in DB)
  • No console errors/warnings
  • Performance acceptable

Available QA Tools

Mobile (React Native/Expo):

  • adb exec-out screencap -p > /tmp/screen.png — Screenshot
  • adb shell am start/force-stop — Launch/restart app
  • ADB tap scripts for UI interaction

Web (Browser):

  • Browser MCP tools (navigate, screenshot, click, fill)
  • DevTools console and network inspection

Backend (Rails/API):

  • Rails console for data verification
  • curl/httpie for API testing
  • Docker logs for error inspection

Human UAT (After Claude QA Complete)

Claude handles ALL testing. Human (Product Manager) does UAT on a thoroughly tested feature:

  • Final product acceptance
  • Subjective UX evaluation
  • Exploratory testing (optional)
  • Sign-off for release

The handoff should be:

"Feature is fully tested. All automated tests pass. I verified the full user flow via [tools used]. Edge cases tested: [list]. Ready for your UAT sign-off."

NOT:

"Tests pass. Please manually verify [list of things Claude should have tested]."

Anti-patterns

You MUST NEVER:

  • Defer "manual" testing to human — use tools instead
  • Hand off with "please verify [X]" when tools could verify it
  • Skip edge case testing because "tests pass"
  • Assume UI works without visual verification via tools
  • Mark complete without verifying data persistence
  • Test only happy path

Claude MUST:

  • Use available tools for visual verification
  • Test with production-like data volumes
  • Verify both UI (via screenshots) and database state
  • Test with multiple user roles/permissions
  • Document unexpected behavior found during testing
  • Hand off with verification evidence

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free