- Per-spec token & cost trackingJSONL
- Phase allocation vs targets
- Per-provider cost breakdown
- Circuit breaker triggers
- Semantic cache hit rate
- Model routing decisions
- Anomaly detection flags
- Auto-calibration profilesNEW
- Lint violations & scoreWEIGHTED
- Type-check errors & score
- Test pass/fail & coverage %
- Quality gate levels
- Baseline regression detection
- Drift direction tracking
- Complexity scoringPLANNED
- Mutation test scorePLANNED
- Layer boundary violationsENFORCED
- Conformance violations
- Import rule enforcement
- Base class compliance
- Required method checks
- Schema-DDL sync
- Coupling analysisPLANNED
- Context window utilization
- File read token estimates
- Cacheable content %
- Narrow scope optimization
- Spec token footprint
- Conversation token tracking
- Stale doc referencesAUTO
- Oversized functions
- Orphaned test files
- Architecture drift
- Budget overrun alerts
- Dead code detection
- Lint auto-fix success rate3-TRY
- Type error remaining count
- Test failure resolution
- Circuit breaker state
- Heal attempt history
- Failure report generation
🛡
Code Health
- Cyclomatic complexity
- Cognitive complexity
- Nesting depth
- Duplication %
- Maintainability index
🧪
Test Quality
- Mutation score
- Branch coverage
- Flaky test detection
- Test-to-code ratio
- Execution time
🔒
Security
- CVE dependency audit
- Static analysis (SAST)
- Secret detection
- License compliance
- OWASP score
🤖
AI-Specific
- Agent edit accuracy
- Rework rate
- First-pass gate rate
- Token efficiency
- Quality per 1K tokens