[FEATURE] repo-compass

[pietrotommasini2022]

Feature Type

• New skill suggestion
• Enhancement to existing skill
• Improvement to documentation
• New installation script feature
• Other

Related Skill (if applicable)

Which skill would this enhance?

• skill-creator
• github-ops
• doc-to-markdown
• mermaid-tools
• statusline-generator
• teams-channel-post-writer
• repomix-unmixer
• llm-icon-finder
• Marketplace infrastructure
• N/A - This is a new skill

Problem Statement

AI assistants (Claude Code, Copilot, etc.) lack codebase context unless they read every file individually — slow, expensive, and token-heavy. There's no standard way to give an AI a dense, structured snapshot
of a repo.

Proposed Solution

A Claude Code skill that runs repo-compass scan to generate CODEBASE.ctx — a compact file (~1000 tokens for a 50-file project) containing exports, imports, module graph, and centrality scores. Claude reads one
file instead of hundreds.

Use Cases

1. Developer opens new repo and wants Claude to understand structure before coding
2. AI-assisted code review needs dependency context without reading every file
3. Onboarding — instant codebase map for new contributors using Claude
   Example Usage

npm install -g repo-compass

then in Claude Code:

/repo-compass

Claude confirms: "This is a TypeScript CLI that generates codebase context..."

Benefits

• 20–50× token reduction vs reading individual files
• No API calls — pure local compute, works offline
• Auto-installs /repo-compass slash command via npm postinstall
• Supports TS, JS (ESM+CJS), Python

Implementation Ideas

Skill already exists on npm (repo-compass) with SKILL.md and .claude/commands/repo-compass.md included in package. Install: npm install -g repo-compass.

Additional Context

GitHub: https://github.com/pietrotommasini2022/repo-compass
npm: https://www.npmjs.com/package/repo-compass

• Benefit 1...
• Benefit 2...
• Benefit 3...

Implementation Ideas

(Optional) Any thoughts on how this could be implemented?

Additional Context

Add any other context, screenshots, or mockups about the feature request here.