Welcome to Cremind

The official documentation for Cremind — the open-source, self-hosted AI agent that reacts to your world in real time.

Cremind is an open-source personal AI agent you run on your own machine or server. It plans and reasons with a powerful model, then carries out work through a unified tool registry that spans built-in tools, Agent Skills, MCP servers, and other agents over A2A — and it reacts to real-world events the moment they happen, instead of polling on a timer.

This portal documents Cremind for two audiences. Pick your path:

I want to run Cremind

Install in one command, finish the browser setup wizard, and start chatting in a few minutes.

I want to build or contribute

Understand the architecture, write Agent Skills, and contribute to the open-source ecosystem.

What makes Cremind different

Two model groups, one focused agent. A high group reasons and plans; a low group runs individual tool calls — so you don't pay frontier-model rates for routine work. See Model Groups.
Real-time, event-driven. Skills push real-world events straight into the agent over a live connection — no cron, no polling. See Event Architecture.
Agent Skills with event listeners. A Skill is a SKILL.md plus a few scripts, and it can ship a long-running listener. See Agent Skills.
Cremind Connect. A token-less, open-source OAuth broker that links Gmail, Calendar, and Jira without you setting up a cloud project — and without storing your data. See Cremind Connect.