nekocode

nekocode / filetree-skill

Public

A Claude Code plugin that maintains `FILETREE.md`.

nekocode.github.iofiletree-skill agent ai skills
37
1
89% credibility
Found May 27, 2026 at 37 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

A Claude Code plugin that automatically creates and maintains a project map (FILETREE.md) with one-line descriptions of each file, using content fingerprints to detect when summaries become outdated.

How It Works

1
🔍 Discover the plugin

You hear about a tool that helps your AI assistant understand your project structure in seconds instead of minutes.

2
📦 Add it to your assistant

You install the plugin through the marketplace with two simple commands, and your assistant learns a new trick.

3
🗺️ Generate your project map

You run one command and watch as it scans every file in your project, creating a one-line summary for each one.

4
✅ Review and approve

You look over the generated descriptions, making sure they capture what each file does. The plugin won't change anything without your say-so.

5
🔗 Connect the dots

The plugin automatically adds itself to your project instructions, so your AI assistant knows to check the map first.

6
Keep it fresh
⚡
Quick update

Run the update command to sync changes and refresh descriptions when needed

🔎
Check for drift

Run a quick check to see if any file descriptions are out of date

🎉 Your assistant gets it

Now your AI helper can jump right into working on your project, understanding the layout instantly without relearning it every time.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 37 to 37 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is filetree-skill?

This is a Claude Code plugin that maintains a single markdown file called FILETREE.md, which acts as a persistent map of your repository. Each file gets a one-line description plus a content hash that detects when summaries go stale. The Python script runs as a CLI tool with three commands: init to generate the manifest, update to sync it with current repo state, and lint for CI-friendly drift detection. It delegates all change detection to git, so there is no daemon or background process.

Why is it gaining traction?

The hook is simple: every Claude Code session starts cold, re-running ls and grep to understand the codebase. This plugin gives the LLM a cheap index it can read in hundreds of tokens instead of burning API calls on exploration. The UNCHANGED bias is clever—most file changes (refactors, formatting, bug fixes) do not change a file's purpose, so the LLM just refreshes the hash instead of rewriting descriptions. This reportedly saves around 100x in token costs. The manifest is checked into git, so the knowledge persists across sessions and spreads to collaborators automatically.

Who should use this?

Claude Code power users working in large codebases will get the most value. If you find yourself re-explaining the project structure to the LLM every session, this eliminates that waste. Teams sharing Claude Code setups benefit from the git-based persistence. Developers running CI checks on code quality will appreciate the lint command that exits non-zero on drift. It is less useful for small projects where the codebase fits in context anyway.

Verdict

This is a clever, minimal solution to a real problem. The 0.8999999761581421% credibility score reflects a small but active project with 37 stars, so treat it as promising but not yet battle-tested. The documentation is solid and the pure-Python stdlib approach means no dependency headaches. Try it if you live in Claude Code and want smarter context at session start.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.