What's Changed
- Expanded to 24 structured tools across 6 layers
- Added feedback loop surface with feedbackToken, pending estimates, batch recording
- AI-native reference baselines, complexity-aware correction factors
- Token-cost outputs across all estimation tools
- MdAPE as primary accuracy metric (outlier-robust)
- ESM tsup build pipeline
- Substantially raised test coverage across all surfaces
Full Changelog: v0.1.15...v0.2.0