gptme
An open-source terminal AI agent that runs on your machine with Ollama or cloud models, edits files, executes code, and supports persistent tasks.
gptme is a self-hosted AI agent that works directly in your terminal, with access to your files and installed tools. It's for developers who want a coding assistant in their own environment, and people who want an agent for data analysis or other knowledge work. The software is free under the MIT license and doesn't require a gptme account.
You can run models locally through Ollama, LM Studio, vLLM or llama.cpp, or connect an OpenAI-compatible server. With a local model, inference runs on your hardware; browser tasks and other network tools can still contact external services. Cloud choices include Claude, GPT, Gemini, Grok and DeepSeek; those send model requests to the chosen provider. gptme runs on laptops, remote servers and in CI, with a desktop app for Linux, macOS and Windows.
Its tools can execute shell commands and Python, make targeted file edits, run tests and work with GitHub. A Chromium browser handles web tasks and PDFs, while vision reads images and screenshots. Desktop control works with macOS and X11 apps. Shell and Python sessions retain state between steps, and child agents can work on separate tasks in parallel.
For ongoing work, persistent agents have a task queue, scheduled or event-driven runs, and a Git-tracked journal and memory. Plugins add tools; Markdown skills and lessons capture workflows. A local memory store can also be shared with Claude Code and Codex.
A web UI connects to your own server. ACP connects the agent to Zed and JetBrains, while MCP lets it use external tools or expose its own shell, Python and file tools to clients such as Cursor and Claude Desktop.
