siorigin

siorigin / atcode

Public

AI code knowledge graph for code understanding, graph retrieval, and agent workflows

19
4
100% credibility
Found Mar 27, 2026 at 19 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

AtCode is an open-source tool that transforms code repositories into searchable knowledge graphs for AI-assisted code exploration, documentation, and integration with coding agents via web UI, API, and MCP.

How It Works

1
🔍 Discover AtCode

You hear about a helpful tool that turns confusing code projects into easy-to-explore maps for AI assistants.

2
🚀 Start with one easy command

Run a simple script that automatically sets everything up, including helpful services, on your computer.

3
🧠 Connect your AI thinking partner

Add a quick setting for an AI service so it can understand and chat about code intelligently.

4
📁 Add your code project

Point to a folder on your computer or a web link, and watch it build a smart map of your code.

5
💬 Chat and explore your code

Ask natural questions in the web page, like 'how does this part work?', and get clear answers with code highlights.

6
Share with AI coding helpers
🔗
Link to desktop AI

Your code map works inside tools like Claude for seamless help.

🌍
Use web or direct access

Keep exploring right in the browser or apps without extra setup.

🎉 Unlock any codebase

Now you effortlessly understand, document, and improve complex projects with AI guidance.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 19 to 19 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 atcode?

AtCode turns GitHub repositories or local codebases into a queryable AST knowledge graph for precise code understanding, dependency tracing, and retrieval. Built in Python with a Next.js frontend and FastAPI backend, it uses Memgraph for storage and supports incremental sync, multi-language parsing (Python strongest), plus paper-to-code workflows from Hugging Face daily papers. Users get a web UI at localhost:3007, HTTP API, and MCP endpoint for agent tools like explore_code, find_nodes, and trace_dependencies.

Why is it gaining traction?

It stands out with MCP integration for code github copilot-like tools (Claude Code, Codex), real-time file watching for code knowledge tracing without full rebuilds, and cross-repo queries for monorepos. Developers notice reliable structural insights—call chains, class hierarchies—over text-based search, plus one-command setup via atcode.sh up handling Docker, uv, and Node deps.

Who should use this?

AI agent builders wiring code knowledge base into workflows, backend devs managing large Python/JavaScript repos needing code knowledge management and tracing, or researchers linking atcoder problems, github python code, and papers for code knowledge reading.

Verdict

Early but capable code knowledge graph github tool with excellent docs, Docker compose options, and MCP hooks—worth prototyping for code github ai tasks. At 19 stars and 1.0% credibility score, validate on your repos first as it's maturing.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.