🦡 codebadger

codebadger is a containerized Model Context Protocol (MCP) server that gives AI agents and LLMs deep, queryable access to a codebase's structure and data flow through Joern Code Property Graphs (CPGs).

Point it at a Git repository, a local path, or even a pasted code snippet, and codebadger builds a CPG and exposes it over MCP — so an assistant can run CPGQL queries, trace data flow and taint, slice programs, and hunt for vulnerabilities across Java, C/C++, JavaScript, Python, Go, Kotlin, C#, Ghidra, Jimple, PHP, Ruby, and Swift.

It's a general-purpose foundation for both program analysis (understanding code structure, call graphs, and data flow) and vulnerability analysis (taint tracking, bug hunting, and PoC development) — useful for academic research as well as industry security and engineering work. It's built to scale to large analysis batches with per-CPG worker pools, memory-aware scheduling, and a Postgres/Redis backend.

News

codebadger and its paper - Bridging Code Property Graphs and Language Models for Program Analysis - were accepted at the Software Vulnerability Management Workshop @ ICSE 2026. 🎉

Related MCP server: CodeFlow MCP Server

Documentation

Everything a developer or security researcher needs lives in docs/:

Found a vulnerability using codebadger?

We'd love to hear about it - open a PR adding it to TROPHIES.md (CVE ID, project, one-line description, date).

Citation

@inproceedings{lekssays2026bridging,
  title={Bridging Code Property Graphs and Language Models for Program Analysis},
  author={Lekssays, Ahmed},
  booktitle={Proceedings of the 2026 IEEE/ACM 4th International Workshop on Software Vulnerability Management},
  pages={33--40},
  year={2026}
}