Our take
- Activity
- 81
- Community
- 93
- Issues
- 69
- Pull requests
- 76
+299 stars in the last 30 days
- Merges more pull requests than 80% of the projects we track
- More contributors than 78% of the projects we track
- Releases more often than 77% of the projects we track
Project history
Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.
Stars
68,897Forks
5,950Change since Nov 16, 2026: +3 stars·+2 forks
Overview
Open Interpreter is a terminal-based coding agent built around a Rust implementation of the Codex agent harness. Its focus is helping users run coding workflows with low-cost models while keeping a familiar agent interface and supporting several model-specific harnesses.
It is intended for developers who want an agent in their existing terminal or editor setup, rather than a standalone product-specific environment. Integrations include Agent Client Protocol (ACP), the Codex exec protocol, MCP, and shared AGENTS.md instructions and skills directories.
Key Features
- Provides a terminal interface for coding-agent sessions and command execution.
- Switches between supported agent harnesses, including Kimi Code, Qwen Code, DeepSeek TUI, and Claude Code variants.
- Connects to ACP-compatible editors and clients through
interpreter acp. - Offers compatibility with the Codex exec protocol and a Codex SDK binary override.
- Supports OpenAI-compatible providers through Chat Completions mode.
- Runs commands with native sandboxing on macOS, Linux, and Windows.
- Includes a QA skill for testing web interfaces with agent-browser and native apps with trycua.
- Uses shared
AGENTS.mdand.agents/skillsconventions for portability across compatible agents.
Use Cases
- Developers can use a terminal agent to edit code and run commands with a supported low-cost model.
- Teams building ACP-compatible editor integrations can run Open Interpreter as the agent backend.
- Existing Codex SDK users can try the implementation by overriding the Codex binary path.
- Developers testing web or native applications can use the included QA skill with the documented companion tools.
Getting Started
Install on macOS or Linux:
curl -fsSL https://www.openinterpreter.com/install | sh
Then run i or interpreter in a terminal. For Windows installation and configuration details, see the README and the official documentation.
Alternatives
- grok-build: Grok Build is a Rust terminal coding agent with web search and CI scripting, while Open Interpreter focuses on low-cost models and ACP or Codex integrations.
- DeepSeek-Reasonix: DeepSeek-Reasonix emphasizes persistent sessions and prefix-cache stability, while Open Interpreter supports multiple agent harnesses and ACP-compatible editors.
- qwen-code: Qwen Code supports multiple model providers and interfaces beyond the terminal, while Open Interpreter targets low-cost models and connects with ACP editors and Codex SDK clients.
- zero: Zero is a Go-based repository agent with local sessions and explicit side-effect controls, while Open Interpreter is Rust-based and supports multiple agent harnesses.
| Project | Language | License | Stars | Status |
|---|---|---|---|---|
| openinterpreter | Rust | Apache-2.0 | 68.5k | Not checked yet |
| grok-build | Rust | Apache-2.0 | 27.3k | New |
| DeepSeek-Reasonix | Go | MIT | 35.9k | Active |
| qwen-code | TypeScript | Apache-2.0 | 28.6k | Active |
| zero | Go | MIT | 1.7k | New |
Considerations
The current repository is the Rust-based Codex-derived version, not the original Python project; the README points to a community-maintained Python fork. Model availability and provider setup depend on provider configuration and, for live model catalogs, provider credentials. Review the sandbox, permissions, and provider documentation before using the agent on sensitive code or systems.
Found this useful?
Share it with someone who would like openinterpreter.