Introducing Nanobot, an Open Source framework for building MCP agents—complete with reasoning, system prompts, tool orchestration, and rich MCP-UI support.
A plain MCP server can expose useful functions. Nanobot takes this further by wrapping your server into an intelligent agent.
Together, the MCP server and Nanobot agent create an interactive experience with a professional dealer persona and rich UI.
Everything you need to build the next generation of AI agents with rich, interactive experiences.
Nanobot was written from the ground up to support the MCP ecosystem. All functionality is delivered via MCP servers, and it fully leverages MCP features like tools, prompts, sampling, elicitation and more.
Nanobots are full AI agents, but they are also exposed as MCP servers so users can chat with them via any app that is an MCP client.
First-class support for MCP-UI spec. Render interactive React components directly inside chat clients—build playable games, shopping carts, dashboards, and more.
Nanobot is a full MCP host with an interactive chat client that can be deployed anywhere, and embedded in any application or website.
Every month, thousands of new MCP servers come online, offering rich functionality that can be added to your Nanobots.
Flexible, extensible, and community-driven. Full transparency with active development and contribution opportunities.
Get up and running with Nanobot in minutes.
Nanobot can be installed via Homebrew:
This will give you the nanobot CLI, which you can use to run and manage your MCP host.
Create a configuration file (e.g. nanobot.yaml) that defines your agents and MCP servers:
The UI will be available at http://localhost:8080.
MCP defined the foundation. Nanobot takes the next step—giving developers the power to create agents that reason, act, and render. Start building today and create the immersive chat experiences you always wanted.
MCP (Model Context Protocol) defines how AI models can interact with external tools and data sources. Nanobot builds on top of MCP by adding agent capabilities—system prompts, conversational memory, autonomous reasoning, and rich UI rendering through MCP-UI.
No! Nanobot is designed to wrap your existing MCP server without requiring changes to your core tools. You simply add agent definitions, system prompts, and UI components on top of your existing functionality.
MCP-UI support is growing across the ecosystem. Nanobot itself includes rich and deep support for MCP-UI, and other clients, such as Goose are adding support for interactive UI components.
Yes! Nanobot is built with production use in mind. It ships as an executable and requires a Postgres database. You can easily run it on any infrastructure platform.