What problem does it solve?
Static analysis tools can drift from ground truth, producing false positives and false negatives that erode trust. This Skill provides a disciplined loop for measuring and improving Fallow's analysis accuracy against competing tools and manually verified source code.
Core Features & Use Cases
- Comparative Benchmarking: Run Fallow and competing analyzers with documented equivalent settings across a stable real-world project corpus.
- Disagreement Classification: Manually verify divergent findings against source and classify each as true positive, false positive, false negative, or model difference.
- Regression-Guarded Fixes: Implement one general correction at a time with a regression fixture, re-run the full corpus, and retain only net improvements.
- Use Case: When Fallow reports unused exports differently than another dead-code tool on a monorepo, use this loop to verify which tool is correct, fix the underlying analysis gap, and lock in the improvement with a fixture.
Quick Start
Run the conformance loop to compare Fallow's dead-code findings against competing tools on a stable corpus and classify every disagreement.
