nicmarti

nicmarti / Claudette

Public

Persistent incremental knowledge graph for token-efficient code reviews with Claude Code

16
3
100% credibility
Found Mar 17, 2026 at 16 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Go
AI Summary

Claudette creates a structural knowledge graph of codebases in supported languages to provide AI coding assistants with focused context for reviews and tasks by tracking only relevant changes.

How It Works

1
🔍 Discover Claudette

You hear about Claudette, a handy helper that lets AI understand your code project by mapping out just the important parts like functions and classes.

2
💻 Set it up on your computer

You quickly prepare Claudette on your machine so it's all set to explore your projects.

3
📁 Add to your project folder

Open your code project and invite Claudette in with a simple one-time setup.

4
🗺️ Build the smart code map

Claudette reads your files and creates a clear picture of how everything connects – now your AI knows exactly where to look!

5
🤖 Connect your AI helper

Restart your AI coding buddy, and it links up with Claudette's map right away.

6
🔄 Watch it update automatically

As you edit code, Claudette keeps the map fresh without you lifting a finger.

🎉 Get perfect code insights

Your AI now reviews changes super smartly, spotting impacts and saving tons of time.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 16 to 16 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 Claudette?

Claudette builds a persistent incremental knowledge graph of your codebase in Go, Python, JS, or TS, parsing structure like functions, classes, imports, and calls. It tracks git changes and feeds Claude Code precise context via an MCP server, slashing tokens by skipping full codebase scans during reviews. Run `claudette build` for initial setup, then `update` or `watch` keeps it fresh automatically.

Why is it gaining traction?

Its single Go binary deploys anywhere with zero deps, unlike LSP tools needing per-language servers. Incremental updates hit under 2s, blast-radius analysis spots impacted files/tests, and MCP tools like `get_impact_radius` or `query_graph` give Claude structural smarts without re-parsing. Persistent github storage ensures the graph survives sessions, echoing persistent systems github reliability.

Who should use this?

Backend devs in Go/Python doing frequent Claude-assisted PR reviews, TS frontend teams chasing callers/callees across modules, or anyone tired of AI hallucinating missing deps. Ideal for medium repos where token costs bite but deep semantics aren't needed.

Verdict

Try it if you're deep in Claude Code workflows—early promise in token savings and git integration shines. With 16 stars and 1.0% credibility, it's raw; solid docs and tests, but watch for lang expansions before production. Worth a `go install` for prototypes.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.