What is Open Interpreter?
Open Interpreter is a coding agent that runs in your terminal. You open a
project, type interpreter, and the agent reads your files, edits them,
and runs commands on your behalf — with sandbox and approval guardrails,
and whatever model you choose.
What it is for
- Making changes to code in an existing project, where the agent needs to read your files to do anything useful.
- Running checks: tests, linters, type-checkers, build scripts. The agent can run them, read the output, and react.
- Automating dev workflows you already do by hand. Pipe it into scripts,
CI, or
githooks with non-interactive mode. - Long coding sessions that need project context. Sessions persist, so
you can pick up tomorrow where you left off with
interpreter resume.
Who it's for
Developers who are comfortable in a terminal. If you already live in
tmux, vim, nvim, zsh, fish, or VS Code's integrated terminal,
Open Interpreter fits where you already work.
How it's built
Open Interpreter is built on top of Codex, the open-source Rust agent core from OpenAI. We took that engine, open-sourced our changes, and made it provider-agnostic: you can plug in OpenAI, Anthropic (directly or via OpenRouter), Kimi, Qwen, DeepSeek, or a fully local model running on your machine.
The Rust core means the TUI is fast, the sandbox is real, and the agent loop is the same battle-tested loop Codex uses.
What it is not for
- General chat questions. If you want a chat box for "what's a closure?", use a chat app. Open Interpreter wants a project directory to work in.
- GUI agent work. If you want an agent that drives your browser, fills forms, reads PDFs, or works across desktop apps, use the Desktop App. The terminal agent stays in the terminal.
See it
$ cd my-project
$ interpreter
> add a /health endpoint that returns the build shaThe agent reads the relevant files, plans the change, shows you the diff, and asks before it touches anything outside the sandbox.