Open source · MIT

Give your coding agent a map of your codebase. 2-4x cheaper, no missed files.

Wingbox builds the code graph of your repository on your machine and tells an AI coding agent, in a few hundred tokens, what a change will break before it edits. Open source (MIT), fourteen languages and frameworks.

$ pipx install wingbox && wingbox install

See it in action

One command, one exact answer. The agent asks which files depend on the one it is about to edit; Wingbox answers from the code graph instead of reading the repository.

A terminal runs wingbox impact on one file of the immich server. On a 3D stack of architecture layers the file lights up in the domain layer, then the ten files that import it in the API, service, domain and test layers, then twenty more at two hops.
Real output of Wingbox 0.2.2 on the server of immich, an open-source photo platform with 707 modules. Checked against the code: an import resolver written separately from Wingbox finds the same 3,795 dependencies, none extra and none missing.

Your architecture, in 3D layers

The build also writes an interactive map: one HTML file that opens in a browser without a network connection. The layers view stacks API, services, domain and infrastructure; click a file and its blast radius lights up across the layers.

One DTO changes: 10 files import it, 30 are within two hops.
Layers view of the Wingbox map with the base service selected: 65 direct dependents, 281 modules within two hops, spread over every layer
The base service changes: 65 direct dependents, 281 of 707 modules within two hops.

What it answers

What could break if I change this file?

Which modules depend on this service, class or type?

Where is this function used, line by line, before I rename it?

Does this dependency cross an architecture boundary?

Where does this new code belong in the architecture?

Which two or three files should I read before fixing this bug?

Run it on your repository

  1. 1$ pipx install wingbox && wingbox install

    On macOS, Linux or Windows: install the command, then put Wingbox in place for every coding assistant on the machine: Claude Code, Codex, Gemini CLI, Cursor and GitHub Copilot. As a plugin: in Claude Code /plugin marketplace add aileron-tech/wingbox, then /plugin install wingbox@aileron-tech; in Codex codex plugin marketplace add aileron-tech/wingbox, then codex plugin add wingbox@aileron-tech. Skill only: npx skills add aileron-tech/wingbox.

  2. 2$ wingbox build .

    Build the graph. Everything Wingbox writes goes into one folder, .wingbox/: the index, the report and the map; wingbox open opens the map. In Claude Code it then keeps itself fresh: a hook marks each edited file, and before the next search only the changed files are patched in, about 1 s for one file on FastAPI. Changes made by git or an editor are found too.

  3. 3$ wingbox impact path/to/file 2

    Ask what a change to one file reaches within two hops.

Benchmarks, a sample report and the live map: wingbox.dev

How it works

  • Everything runs on your machine: no LLM call, no embeddings, no network access to build or query the graph
  • Deterministic static parsing: the same code gives the same graph
  • Python, TypeScript, Rust, C#, Java, Go, Ruby, Rails, PHP, Laravel, Dart, Flutter, Free Pascal and Lazarus
  • An agent skill for Claude Code, Codex, Gemini CLI, Cursor and GitHub Copilot, an MCP server for any MCP client, and a plain command for people

Wingbox is the engine behind Agent Control, our fixed-price audit of agent-written code.