What criteria are used to score the quality of a CLAUDE.md file?

Files are evaluated based on five main criteria: command accuracy, architecture clarity, documentation of non-obvious patterns, conciseness, and actionability.

Can I use this for local settings that I don't want to commit to Git?

Absolutely. The skill recognizes and can manage .claude.local.md files, which are intended for personal preferences and local environment overrides.

Does it support monorepos?

Yes, it is specifically configured to scan for nested CLAUDE.md files within packages or subdirectories, ensuring that module-specific context is maintained alongside the project root.

Does this skill automatically overwrite my files?

No. The skill is designed to find and audit files first, providing a detailed quality report. It only applies targeted updates after you have reviewed the proposed changes and given explicit approval.

CLAUDE.md Context Improver

Name: CLAUDE.md Context Improver
Author: SwervoJake

bySwervoJake

0•

コンテンツ管理

Audits, evaluates, and optimizes CLAUDE.md files to ensure Claude Code maintains the most accurate and actionable project context.

This skill streamlines the maintenance of project context by scanning repositories for CLAUDE.md files—including local overrides and package-specific versions—and evaluating them against professional documentation standards. It generates a comprehensive quality report that scores files on architecture clarity, command accuracy, and actionability. After a human review, the skill applies targeted updates to fix stale build scripts, document non-obvious code patterns, and remove verbose explanations, ensuring your AI development environment is always grounded in the current state of your codebase.

主な機能

01Metric-based quality assessment scoring from A to F

02Identification of stale commands, missing dependencies, and undocumented gotchas

030 GitHub stars

04Automated discovery of root, local, global, and monorepo CLAUDE.md files

05Detailed reporting on architecture clarity and command actionability

06Interactive update process with git-style diffs for user approval

ユースケース

01Optimizing project context to reduce AI hallucinations and command errors

02Maintaining documentation consistency across complex monorepo structures

03Auditing a legacy repository to generate a fresh, accurate CLAUDE.md for AI assistance

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add swervojake/claaaude-studio claude-md-improver

For use in Claude.ai and ChatGPT

主な機能

01Metric-based quality assessment scoring from A to F

02Identification of stale commands, missing dependencies, and undocumented gotchas

030 GitHub stars

04Automated discovery of root, local, global, and monorepo CLAUDE.md files

05Detailed reporting on architecture clarity and command actionability

06Interactive update process with git-style diffs for user approval

ユースケース

01Optimizing project context to reduce AI hallucinations and command errors

02Maintaining documentation consistency across complex monorepo structures

03Auditing a legacy repository to generate a fresh, accurate CLAUDE.md for AI assistance

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add swervojake/claaaude-studio claude-md-improver

For use in Claude.ai and ChatGPT