| name | maestro-qa-report |
| description | Use when a React Native or Expo app needs report-only mobile QA with Maestro on a simulator or emulator. Verifies local prerequisites, runs flows with explicit artifact and report output, captures screenshots and logs, classifies severity, and produces exact repro notes without making code changes. |
| metadata | {"tags":"maestro,mobile,qa","groups":"mobile-qa","invocation":"model"} |
Maestro QA Report
Overview
Run Maestro as a report-only QA pass.
This skill is for verifying behavior and documenting failures, not fixing them. Use it before a fix-capable QA skill, before release-readiness work, or when the user wants reproducible mobile QA evidence from a local simulator or emulator run.
Default workflow
1. Confirm the test surface
Before running anything, establish:
- target platform: iOS simulator or Android emulator
- app type: Expo or bare React Native
- app binary: development build, preview build, or simulator/emulator build
- flow location: usually
.maestro/ in the app repo
For Expo projects, prefer a development build or dedicated test build. Do not treat Expo Go as the default serious QA target.
2. Verify prerequisites
Check for:
maestro installed- a running simulator or emulator
- flow files present
- a stable output directory for artifacts
Use the bundled helper:
skills/maestro-qa-report/scripts/run-maestro-qa-report.sh \
--repo /path/to/app \
--flow-path .maestro \
--output-dir build/maestro-results \
--report-format html-detailed \
--report-file build/maestro-report.html
If the repo uses tags, pass them through:
skills/maestro-qa-report/scripts/run-maestro-qa-report.sh \
--repo /path/to/app \
--flow-path .maestro \
--include-tags smoke
3. Run flows and capture artifacts
Prefer:
- short smoke flows first
- explicit output paths
- HTML or HTML-detailed reports for human review
- a separate debug directory when logs matter
Current Maestro guidance supports:
--test-output-dir for screenshots, videos, commands JSON, and related artifacts--format junit|html|html-detailed--output for report file paths
4. Turn failures into QA findings
For each verified failure, record:
- severity:
blocker, major, minor, or cosmetic
- exact flow and step
- exact repro path
- expected vs actual behavior
- artifact links or paths
- suspected ownership: app code, test flow, seed data, environment, or backend dependency
Use the report template in references/report-template.md.
5. Do not fix code in this skill
This is report-only.
Do not:
- edit app code
- rewrite flows unless the failure is clearly a broken test artifact and the user asked for that
- collapse environment failures and product bugs into one bucket
Escalate with a clean report instead.
Severity rules
| Severity | Meaning |
|---|
blocker | cannot complete a critical user path or app cannot be meaningfully tested |
major | core flow works poorly or fails in a release-significant way |
minor | non-critical behavior is broken but core flow still works |
cosmetic | copy, spacing, visual, or low-risk polish issue |
When uncertain, prefer major over blocker unless the app is unusable for the target flow.
Quick reference
| Need | Action |
|---|
| run local report-only QA | use the helper with explicit output paths |
| keep artifacts organized | set --output-dir and --report-file |
| narrow run scope | use --include-tags |
| debug environment issues | add --debug-output-dir |
| write the final summary | use the report template |
Common mistakes
- running against Expo Go when the issue depends on a real dev or preview build
- treating missing simulator setup as an app bug
- running a giant suite before proving one smoke flow works
- reporting a failure without artifact paths or exact flow steps
- fixing code during a report-only QA pass
Example
If the user asks:
run maestro qa and give me a report for the login and onboarding flows
do this:
- verify Maestro, device, and flow files
- run the targeted flows with explicit report and artifact output paths
- inspect the generated report plus screenshots and logs
- write a concise QA report with severity, repro steps, and suspected ownership
- stop at the report unless the user explicitly asks to fix issues
