proto is a multi-model AI agent for the terminal. Connect it to any OpenAI-compatible, Anthropic, or Gemini endpoint and let it read, write, and run code directly in your project.
| I want to… | Go to… |
|---|---|
| Install proto and run my first session | Tutorials → Getting Started |
| Accomplish a specific task (MCP, hooks, headless, etc.) | Guides |
| Look up settings, commands, or tool signatures | Reference |
| Understand how proto works under the hood | Explanation |
| Contribute to the codebase or build on the SDK | Contributing |
Step-by-step walkthroughs for learning proto from scratch. Follow along to install, configure, and run your first agentic sessions.
Task-oriented how-tos for specific goals — connecting models, writing sub-agents, using hooks, running in CI, and more.
Exact, complete reference material: all settings keys, slash commands, keyboard shortcuts, tool signatures, and the beads task tracker CLI.
Conceptual documentation explaining how proto's subsystems work and why they are designed the way they are.
Architecture deep-dives, development workflow, SDK guides, and examples for contributors and extension builders.