What it does

brooks-lint is a set of Agent Skills for AI coding assistants. Rather than counting lines or complexity, it asks the assistant to diagnose code against six decay risks, such as cognitive overload, change propagation and knowledge duplication, drawing on twelve named engineering books. Each finding comes back in a fixed Symptom, Source, Consequence, Remedy shape with a book citation and a severity label, and the report ends with a health score out of 100. Six commands cover pull-request review, architecture audit with a dependency graph, tech-debt assessment, test-suite review, a quick health check, and a sweep that applies safe fixes automatically and flags architectural decisions for a person. It installs as a Claude Code plugin or through an install script for other agent platforms, and an optional .brooks-lint.yaml file in the project root sets strictness and can disable individual checks. The README says it does not replace a linter. One caveat from the maintainer: a note dated 2026-10-16 says only Codex-related capabilities will be tested going forward, so behaviour on Claude Code, Gemini CLI and other platforms is not currently verified by the maintainer.

Capabilities

Six decay-risk lenses for production code and six for test suites
Findings with book citations, severity labels and a suggested remedy
Health score from 0 to 100 with configurable strictness
Architecture audit with a colour-coded Mermaid dependency graph
Sweep mode that applies safe fixes and flags architectural ones for review
Installs on Claude Code and a long list of other Agent Skills platforms

When to use it

You want AI code reviews that follow a consistent structure and cite their reasoning
You need a quick read on architecture drift or tech debt before a release
You want to audit the quality of a test suite as well as the production code