Generate context-aware quality checklists for code review and QA using IEEE 1028 base standards plus LLM contextual additions
version
1.1.0
model
sonnet
invoked_by
both
user_invocable
true
tools
["Read","Write","Edit","Glob","Grep"]
best_practices
["Start with IEEE 1028 base checklist (80-90% coverage)","Add contextual items based on project type (10-20%)","Mark AI-generated items with [AI-GENERATED] prefix","Tailor checklist to current file types and frameworks","Return markdown checklist with checkboxes"]
# Quality Checklist
Generated: 2026-01-28 10:30:00
Context: TypeScript, React, REST API
## Code Quality (IEEE 1028)- [ ] Code follows project style guide
- [ ] No code duplication
- [ ] Cyclomatic complexity < 10
- [ ] Functions have single responsibility
- [ ] Variable names clear and descriptive
- [ ] Magic numbers replaced with constants
- [ ] Dead code removed
## Testing (IEEE 1028)- [ ] Tests written first (TDD)
- [ ] All new code has tests
- [ ] Tests cover edge cases
- [ ] Test coverage ≥ 80%
- [ ] Integration tests present
- [ ] Tests isolated (no order dependency)
## Security (IEEE 1028)- [ ] Input validation on all inputs
- [ ] No SQL injection risks
- [ ] No XSS vulnerabilities
- [ ] Sensitive data encrypted
- [ ] Auth/authz checks present
- [ ] No hardcoded secrets
- [ ] OWASP Top 10 reviewed
## Performance (IEEE 1028)- [ ] No performance bottlenecks
- [ ] Database queries optimized
- [ ] Caching used appropriately
- [ ] Resource cleanup (connections)
- [ ] No infinite loop risks
- [ ] Large data paginated
## Documentation (IEEE 1028)- [ ] Public APIs documented
- [ ] Complex logic has comments
- [ ] README updated
- [ ] CHANGELOG updated
- [ ] Breaking changes documented
## Error Handling (IEEE 1028)- [ ] All errors handled
- [ ] User-friendly error messages
- [ ] Detailed logs for debugging
- [ ] No swallowed exceptions
- [ ] Graceful degradation
- [ ] Rollback procedures
## TypeScript (AI-Generated)- [ ] [AI-GENERATED] Types exported from modules
- [ ] [AI-GENERATED] No `any` types (justified if used)
- [ ] [AI-GENERATED] Strict null checks satisfied
- [ ] [AI-GENERATED] Interfaces prefer over types
## React (AI-GENERATED)- [ ] [AI-GENERATED] Components use React.memo appropriately
- [ ] [AI-GENERATED] Hooks follow Rules of Hooks
- [ ] [AI-GENERATED] No unnecessary re-renders
- [ ] [AI-GENERATED] Keys on list items
## REST API (AI-GENERATED)- [ ] [AI-GENERATED] Rate limiting implemented
- [ ] [AI-GENERATED] API versioning in URLs
- [ ] [AI-GENERATED] Request/response validation
- [ ] [AI-GENERATED] OpenAPI/Swagger updated
---
**Total Items**: 38
**IEEE Base**: 30 (79%)
**Contextual**: 8 (21%)
Best Practices
DO
Start with IEEE 1028 base (universal quality)
Analyze project context before generating
Mark all LLM items with [AI-GENERATED]
Keep contextual items focused (10-20%)
Return actionable checklist (not generic advice)
DON'T
Generate checklist without context analysis
Exceed 20% contextual items (dilutes IEEE base)
Forget [AI-GENERATED] prefix
Include items that can't be verified
Make checklist too long (>50 items)
Iron Laws
NEVER complete a task without checklist validation — use verification-before-completion to enforce this gate; skipping checklist validation defeats the entire purpose of the skill.
ALWAYS analyze project context before generating the checklist — detect framework, language, and patterns first; a generic checklist without context misses stack-specific items and includes irrelevant ones.
NEVER exceed 50 checklist items — checklists longer than 50 items are not completed in practice; be ruthlessly selective and prioritize blocking items over nice-to-haves.
ALWAYS mark LLM-generated items with [AI-GENERATED] prefix — IEEE 1028 base items are verified standards; AI-generated items are contextual suggestions requiring human judgment.
NEVER include non-verifiable items — every checklist item must be objectively checkable with a yes/no answer; vague items like "code is clean" cannot be definitively validated.
Anti-Patterns
Anti-Pattern
Why It Fails
Correct Approach
Generating without context analysis
Produces generic, irrelevant checklist items
Detect framework/language/patterns first
Exceeding 50 items
Long checklists are abandoned mid-review
Cut ruthlessly; keep to ≤50 high-value items
No [AI-GENERATED] prefix on contextual items
Conflates standards with suggestions
Mark all AI-contextual items clearly
Non-verifiable items ("code is clean")
Can't be definitively checked off
Only include objectively pass/fail items
Skipping checklist for "small" changes
Small changes introduce large bugs
Apply checklist proportionally for every change
Related Skills
qa-workflow - Systematic QA validation with fix loops