gptme

Active
GitHub Python MIT

Description

gptme is a terminal-based personal AI agent that can run shell commands, read and write files and drive a browser to complete real development tasks.

Key Features

  • Terminal-native — runs directly in your shell, no extra UI to install
  • Real tool use — built-in shell, file I/O and browser tools so the agent can actually do work
  • Local-first — runs locally by default and works with any OpenAI-compatible LLM
  • Extensible tools — register custom tools via Python modules to grow agent capabilities
  • Persistent sessions — history replay and saved context for reproducibility and auditing
  • Server mode — ships gptme-server for Web and API access

Use Cases

💡 Letting an agent write code, fix bugs and run tests from your terminal
💡 Letting the agent drive a browser to operate web apps automatically
💡 Embedding gptme inside a local IDE or SSH workflow
💡 Auditing the agent's real actions via persistent session logs
💡 Running gptme-server to provide agent services to internal teams

Tags

coding-agent terminal cli agent python

Categories

💬 Chat UI 💻 Coding Agent

Quick Start

# Install via pipx (recommended)
pipx install gptme

# Or via pip
pip install gptme

# Configure the LLM
export OPENAI_API_KEY=sk-xxx

# Start an interactive session
gptme 'Create a hello.py in this directory that prints Hello, gptme!'

# Launch the web server
gptme-server
Visit GitHub Visit Website View Docs

Related Projects