Brooks-Lint

What's New in v0.4.0

Framework Redesign

• Six-book framework: Replaced eight Brooks-only dimensions with six cross-book decay risk dimensions synthesized from The Mythical Man-Month, Code Complete, Refactoring, Clean Architecture, The Pragmatic Programmer, and Domain-Driven Design
• Diagnosis chain: Every finding now follows Symptom → Source → Consequence → Remedy (the Iron Law)
• Severity labels: 🔴 Critical / 🟡 Warning / 🟢 Suggestion replace numeric scoring
• Health Score: Deduction-based (100 − 15×Critical − 5×Warning − 1×Suggestion)

New Files

• decay-risks.md — Six decay risks with full symptom lists and source attributions
• evals/evals.json — Benchmark test suite (3 real-world scenarios)
• CONTRIBUTING.md — Three contribution paths, local testing guide, PR conventions

Rewrites

• SKILL.md — Iron Law enforcement, mode detection, new report template
• pr-review-guide.md — Six-risk scan ordered by PR relevance
• architecture-guide.md — Six-risk audit + Conway's Law check
• debt-guide.md — Pain×Spread priority formula + Debt Summary Table
• README.md — Benchmark data (94% vs 16%), real output showcase, comparison table

Benchmark

Tested across 3 real-world code scenarios:

| Criterion | brooks-lint | Claude alone |
|-----------|:-----------:|:------------:|
| Structured findings (Symptom → Source → Consequence → Remedy) | ✅ 100% | ❌ 0% |
| Book citations per finding | ✅ 100% | ❌ 0% |
| Severity labels (🔴/🟡/🟢) | ✅ 100% | ❌ 0% |
| Health Score (0–100) | ✅ 100% | ❌ 0% |
| Overall pass rate | 94% | 16% |