Repolode
haseeb-heaven

haseeb-heaven/open-agent

Terminal AI agent that plans and runs tasks using free, local (Ollama, LM Studio) or bring-your-own-key cloud models.

Open-source alternative to:Claude CodeOpen Interpreter

Stars
278
Forks
43
Open issues
7
Last commit
last month

What is it?

OpenAgent is an open-source AI agent you run from the terminal. You describe a task in plain English, and it plans steps, calls tools, and produces results, such as analyzing a CSV file and plotting a chart. It works with free OpenRouter models, local runtimes like Ollama and LM Studio, and a range of cloud providers including OpenAI, Anthropic, Gemini, Groq, and DeepSeek, using your own API keys. No account is required.

Model selection is handled through a registry file and CLI flags: you can pick a provider and model explicitly, use a free rotation with automatic fallback, or switch models inside a session. Web search picks a backend based on which keys are available, with a no-key fallback chain through Exa and DuckDuckGo. Extensions can be installed from a marketplace or local manifests, including Claude-style plugin manifests and MCP server declarations.

The project is published on npm and requires Node.js 22 or later. It is a fork of Google's Gemini CLI, licensed under Apache-2.0, and runs on Windows, macOS, and Linux.

Key features

  • 01Runs tasks described in plain English, including planning and tool use.
  • 02Supports free OpenRouter models with automatic fallback.
  • 03Connects to local models through Ollama and LM Studio without an API key.
  • 04Works with many cloud providers using your own keys via a .env file or interactive setup.
  • 05Auto-selects a web search backend, with a no-key fallback to Exa and DuckDuckGo.
  • 06Installs extensions from manifests, including MCP server declarations.

Use cases

  • Analyzing a local CSV or spreadsheet and generating charts from the terminal.
  • Experimenting with open-source or local models without paying for API access.
  • Running an agent against a local Ollama setup for privacy-sensitive work.
  • Comparing multiple LLM providers from one command-line interface.

Quick start

  1. 1

    Install from npm

    npm install -g @haseeb_heaven/open-agent
  2. 2

    Launch the agent

    openagent

Things to know

  • Requires Node.js 22 or newer.
  • Cloud models need your own API key; local models need Ollama or LM Studio running.
  • Latest multi-provider work lives on the feature branch rather than the default branch.

In collections

Topics

FAQ

What is open-agent?
Terminal AI agent that plans and runs tasks using free, local (Ollama, LM Studio) or bring-your-own-key cloud models. OpenAgent is an open-source AI agent you run from the terminal. You describe a task in plain English, and it plans steps, calls tools, and produces results, such as analyzing a CSV file and plotting a chart. It works with free OpenRouter models, local runtimes like Ollama and LM Studio, and a range of cloud providers including OpenAI, Anthropic, Gemini, Groq, and DeepSeek, using your own API keys. No account is required.
Is open-agent free to use? What license does it have?
open-agent is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is open-agent still maintained?
Updated within 6 months. The last commit to open-agent was on Oct 6, 2026.
What can I use open-agent for?
Analyzing a local CSV or spreadsheet and generating charts from the terminal. Experimenting with open-source or local models without paying for API access. Running an agent against a local Ollama setup for privacy-sensitive work. Comparing multiple LLM providers from one command-line interface.
How do I install open-agent?
Install from npm: npm install -g @haseeb_heaven/open-agent Launch the agent: openagent
Where can I get open-agent?
The source code and installation instructions are on GitHub: https://github.com/haseeb-heaven/open-agent

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects