ahmed3elshaer/everything-claude-code-mobile/mobile-verification
Claude Code and agent skill by ahmed3elshaer/everything-claude-code-mobile
- community
Checking that the SKILL.md file is still in the repository.
Automated testing workflow with pass@k metrics for mobile development. Detects flaky tests and ensures code reliability.
Install
npx vskill@latest install ahmed3elshaer/everything-claude-code-mobile/mobile-verificationSKILL.md
Mobile Verification Skill
Comprehensive testing workflow with pass@k metrics for Android development reliability.
Philosophy
Single test runs lie.
A test that passes once might fail tomorrow. Verification loops run tests multiple times to reveal:
- Flaky tests (timing issues, async problems)
- Intermittent failures (resource contention)
- Reliability trends (improving vs degrading)
Pass@k Explained
Pass@k = proportion of test iterations that passed
Verification Levels
Quick Verification (k=2)
Standard Verification (k=3)
Thorough Verification (k=5)
Release Verification (k=10)
Test Type Strategies
Unit Tests (JUnit)
Characteristics:
- Fast: ~1-2 seconds per test
- Isolated: No Android dependencies
- Reliable: Should be Pass@k = 1.0
Target Pass@k: ≥ 0.95 (95%)
Common Flaky Causes:
- Async operations without proper waiting
- Date/time dependencies
- Random data generation
- Static state leakage
Fix Strategies:
UI Tests (Espresso)
Characteristics:
- Slow: ~5-10 seconds per test
- Device-dependent: Need emulator/device
- Fragile: UI changes break tests
Target Pass@k: ≥ 0.80 (80%)
Common Flaky Causes:
- Idling resource not registered
- Animation interference
- Screen rotation
- Network timeouts
Fix Strategies:
Compose Tests
Characteristics:
- Fast: ~1-3 seconds per test
- UI-level: Tests Composable behavior
- Modern: Uses Compose Testing framework
Target Pass@k: ≥ 0.90 (90%)
Common Flaky Causes:
- Recomposition timing
- State hoisting issues
- Animation interference
Fix Strategies:
Verification Workflow
During Development
Before Commit
Before Push
Before Release
Interpreting Results
Pass@k Scores
| Score | Meaning | Action | |
|---|---|---|---|
| 1.0 | Perfect | Celebrate | |
| 0.8-0.9 | Excellent | Monitor | |
| 0.6-0.7 | Good | Investigate | |
| 0.4-0.5 | Fair | Fix needed | |
| 0.0-0.3 | Poor | Block release |
Trends
Track pass@k over time:
Flaky Test Patterns
| Pattern | Likely Cause | |
|---|---|---|
| Fails on iteration 1 only | Cold start issue | |
| Fails randomly | Async timing | |
| Fails on specific iteration | Resource leak | |
| Fails in parallel only | Shared state |
Fixing Flaky Tests
Step 1: Identify Pattern
Look for patterns in failures.
Step 2: Add Diagnostics
Step 3: Apply Fix
Common fixes:
- Add
advanceUntilIdle()for coroutines - Add
IdlingResourcefor network - Disable animations for UI tests
- Use
@UiThreadTestfor main thread work - Add explicit waits for async operations
Step 4: Verify Fix
Target: Pass@5 = 1.0
Integration
With Checkpoints
Create checkpoint before verification:
With Memory
Track pass@k in memory:
With Instincts
Learn testing patterns:
Thresholds by Context
| Context | Pass@k Threshold | Rationale | |
|---|---|---|---|
| Unit tests | 0.95 | Should be deterministic | |
| UI tests | 0.80 | More fragile, device-dependent | |
| Compose tests | 0.90 | Better than Espresso, more stable | |
| Integration tests | 0.70 | Complex, more variables | |
| E2E tests | 0.60 | Full system, many variables |
Best Practices
- Start High, Go Low: Use k=5 for investigation, k=3 for routine
- Fix Flaky Fast: Don't tolerate flaky tests
- Track Trends: Monitor pass@k over time
- Context Matters: UI tests can have lower thresholds than unit
- Block Release: Failed verification should block releases
Remember: A test that sometimes passes is worse than no test at all. It gives false confidence.
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/mobile-verification/SKILL.md
Popularity
- Repository stars
- 32
- Repository forks
- 3
- 7-day trend score
- 17
Stars and forks are for the whole GitHub repository, not this skill alone. Refreshed 16 Mar 2026, 7 months ago.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/ahmed3elshaer/everything-claude-code-mobile/mobile-verification)