| name | test-strategy |
| description | Project-specific test strategy. Defines what to verify beyond generic
dogfooding: domain-specific checks, content rendering, build validation,
and error boundaries. Use alongside the dogfood skill which provides
the methodology.
|
| allowed-tools | ["Bash","Read","Grep","Glob"] |
Test Strategy
What to check on this project, on top of dogfood's generic exploratory testing.
This skill defines the WHAT. The dogfood skill defines the HOW.
Stack context
{CONFIGURE: Stack context — tech stack, routing structure, content architecture}
Checklist
1. {CONFIGURE: Test category name}
- {CONFIGURE: Check description}
- {CONFIGURE: Check description}
2. {CONFIGURE: Test category name}
- {CONFIGURE: Check description}
3. Network health
At each page, inspect network requests:
agent-browser network requests --status 4xx --session {SESSION}
agent-browser network requests --status 5xx --session {SESSION}
Flag:
- Failed requests — any 4xx/5xx responses (broken assets, failed API calls)
- Mixed content — HTTP resources loaded on HTTPS page
- Unexpected third-party calls — domains that shouldn't be there
- Slow resources — anything visibly blocking page render
4. Content rendering
- Index pages show item lists, not empty
- Detail pages render body content with headings
- Homepage hero, CTAs, and all sections load
5. Error boundary
/nonexistent-page renders custom 404 with working nav back to home- No white screen or crash
6. Responsive
Test at desktop (1280x720) and mobile (375x812):
- No horizontal overflow
- Text readable, not clipped
- Mobile navigation opens and contains correct links
- CTAs tappable on mobile
7. Build validation
Run before or after browser testing to catch non-visual issues:
- Type checks: {CONFIGURE: Type check command}
- Linting: {CONFIGURE: Lint command}
- Tests: {CONFIGURE: Test command}
- Production build: {CONFIGURE: Build command}
Report pass/fail for each. If any check fails, diagnose but do NOT auto-fix.
Gotchas
{CONFIGURE: Known gotchas and edge cases}
