Self-hostable personal AI assistant as a single Go binary, combining LLM chat, RAG knowledge bases, and agent loops for browsing, shell, and MCP tools.
- Stars
- 6,836
- Forks
- 883
- Open issues
- 6
- Last commit
- yesterday
What is it?
Open Duet is an open-source workspace for creating named AI agents, each with its own instructions, model, workspace folder, memory, and approval settings. A lead agent can delegate tasks to specialist agents, and agents can keep notes and recall earlier conversations. Routines can run prompts on a schedule, through a webhook, or on demand.
The app runs on your own computer or server and does not include a model or model subscription. You connect an engine you already use, such as Claude Code, Codex CLI, Gemini CLI, OpenRouter, or an OpenAI-compatible API like Ollama or LM Studio. Users choose between Ask first, Accept edits, and Plan only modes to control what agents can do, and tool activity is visible in the app.
The project is in active development. The desktop app, server mode, optional Docker agent computers, and team delegation are still evolving, and team leads currently need to use Claude.
Key features
- 01Create named agents with their own instructions, model, workspace, and memory.
- 02Delegate work from a lead agent to specialist agents and follow results in chat.
- 03Schedule routines by time, webhook, or manual run, with approval still required.
- 04Choose Ask first, Accept edits, or Plan only approval modes.
- 05Connect CLI engines, OpenRouter, or OpenAI-compatible APIs such as Ollama.
- 06Store conversations and settings locally or on a self-hosted server.
Use cases
- Running a researcher and an editor on a shared project folder to prepare regular briefings.
- Keeping a set of specialist agents for writing, planning, or support with distinct instructions.
- Automating a recurring prompt on a server while your computer is off.
- Using a local model through Ollama or LM Studio while keeping data on your own machine.
In collections
Topics
FAQ
- What is open-duet?
- Self-hosted workspace for creating named AI coworkers with their own instructions, memory, routines, and approval controls. Open Duet is an open-source workspace for creating named AI agents, each with its own instructions, model, workspace folder, memory, and approval settings. A lead agent can delegate tasks to specialist agents, and agents can keep notes and recall earlier conversations. Routines can run prompts on a schedule, through a webhook, or on demand.
- Is open-duet free to use? What license does it have?
- open-duet is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
- Is open-duet still maintained?
- Actively maintained. The last commit to open-duet was on Nov 6, 2026.
- What can I use open-duet for?
- Running a researcher and an editor on a shared project folder to prepare regular briefings. Keeping a set of specialist agents for writing, planning, or support with distinct instructions. Automating a recurring prompt on a server while your computer is off. Using a local model through Ollama or LM Studio while keeping data on your own machine.
- Where can I get open-duet?
- The source code and installation instructions are on GitHub: https://github.com/Anil-matcha/open-duet
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
A self-hosted personal AI agent server in a single Rust binary, with sandboxed tool execution, multiple LLM providers, and chat channels like Telegram and Discord.
Self-hosted personal AI agent framework in Python with a browser WebUI, terminal client, chat app channels, tools, memory and MCP support.
Open-source personal AI assistant and Agent Harness that plans tasks, runs tools and skills, and keeps memory across web and IM channels.
A self-improving AI agent that creates and refines skills from experience, runs on many model providers, and works through CLI and messaging apps.
A self-hosted Go AI agent that runs 24/7 on your own computer, auto-builds missing tools, and shares them with Claude Code and other agents via MCP.