hyhmrright/brooks-lint v1.0.0

v1.0.0 — Health Dashboard, GitHub Action & Eval Automation

on GitHub

What's New in v1.0.0

This is the first stable release of brooks-lint, bringing ten new features built on top of the v0.9.x foundation.

✨ New Features

F1 — Step Numbering Validation

npm run validate now automatically checks that each SKILL.md Process section's step list matches the corresponding guide's step count and numbering. No more silent drift when adding steps to a guide.

F2 — Auto Scope Detection

All five skills automatically detect the review scope when no files or directory are specified — using git diff, recent commits, or directory heuristics. No more "please point me to a file" friction.

F3 — Health Dashboard (/brooks-health)

A new fifth skill that runs abbreviated scans across all four quality dimensions and produces a weighted composite Health Score. Use it before a release, when onboarding a new team, or whenever you want a big-picture "how are we doing?" report.

F4 — Health Score Trend Tracking

scripts/history.mjs appends each Health Score run to .brooks-lint-history.json. The CI reviewer and health dashboard can display a trend line (↑ / ↓ / →) alongside the current score.

F5 — Post-Report Triage Mode

After any review, teams can request a triage-focused follow-up that prioritizes findings by effort vs. impact and groups them into sprint-sized batches. Suppression config example added to .brooks-lint.example.yaml.

F6 — Fix Mode (--fix)

Any skill can be run with --fix to request concrete, copy-paste-ready remedies (code snippets, refactoring steps) for each finding, rather than general guidance.

F7 — Team Onboarding Report

brooks-audit --onboarding produces a new-developer orientation report: territory map, reading-order dependency graph, conventions, danger zones, domain glossary, and suggested first tasks. No Health Score or Iron Law format — this mode explains, not diagnoses.

F8 — GitHub Action

.github/actions/brooks-lint/action.yml provides a ready-to-use composite action that runs a PR review on every pull request and posts the report as a PR comment. See docs/github-action-example.yml for a copy-paste workflow.

F9 — Live Eval Runner

scripts/run-evals-live.mjs executes all 49 benchmark scenarios against the Claude API and reports per-risk accuracy (e.g., R1: 6/7 (85%)). Supports --id, --risk, --mode, and --model flags for targeted runs. Requires ANTHROPIC_API_KEY.

F10 — Custom Decay Risk Extension

Teams can define project-specific Cx risk codes in .brooks-lint.yaml. Custom risks are scanned, cited with Iron Law format, and included in the Health Score. See skills/_shared/custom-risks-guide.md for the template.

🔧 Improvements

• Skill descriptions — all five SKILL.md descriptions now lead with purpose and concrete trigger phrases; each has a strengthened "Do NOT trigger for:" guardrail
• Eval classification — no_risk_codes and no_health_score boolean flags replace fragile name-substring detection; flags are mutually exclusive and validated by npm run evals
• gemini-extension.json — added missing brooks-health skill and command entries; all four platform manifests now consistently list five analysis modes

🐛 Fixes

• Removed stale VS Code references from documentation; VS Code Extension is out of scope
• Fixed GitHub Action cache pattern (proper actions/cache@v4 hit guard, removed redundant directory check)
• Removed dead code branch for no_health_score name-substring detection in live eval runner

📦 Install / Upgrade

# Via plugin marketplace (recommended)
/plugin marketplace add hyhmrright/brooks-lint
/plugin install brooks-lint@brooks-lint-marketplace

# Manual update
cp -r skills/ ~/.claude/skills/brooks-lint

📚 Available Skills