CodeHUD

CodeHUD is a Rust tool for analyzing large codebases. It builds call graphs, dependency maps, and topology/quality/security summaries that are useful both to human developers and to LLM coding assistants, the main use case is giving an AI agent (e.g. Claude Code) a structural map of a project so it doesn't have to grep everything to understand it.

It supports Rust, Python, JavaScript/TypeScript, Java, C/C++, Go, Ruby, PHP, C#, Swift, and Kotlin via tree-sitter, with per-language queries in queries/. Text summaries can be generated with a local LLM (Ollama) or the Gemini API.

This started as a Python prototype and is a first Rust project, built over several months. It works, but expect rough edges.

Workspace layout

Build

cargo build --release

Requires a recent stable Rust toolchain. Graphviz is needed for rendering call graphs (sudo apt install graphviz / pacman -S graphviz / brew install graphviz).

Usage

# Analyze a codebase (views: topology, quality, security, dependencies)
cargo run --release --bin codehud -- analyze /path/to/codebase --view topology --output analysis.json

# Generate call graph visualizations (overview, per-module, cycle detection)
cargo run --release --bin codehud -- call-graph /path/to/codebase -o my_graph -f svg

See Docs/ for more detailed documentation.

License

MIT, see LICENSE.