AGiXT is a free, open-source AI agent automation platform that lets you build agents able to plan and execute complex, multi-step tasks across many different AI providers. It combines adaptive memory, a plugin system of extensions and chained instructions, so a single agent can browse, write code, manage files and call external services to get work done autonomously.
AGiXT is an open-source project on GitHub with about 3,200 stars, under active development in 2026. It is provider-agnostic: agents can run on OpenAI, Anthropic, local models via llama.cpp and many other backends, and instructions can be chained into reusable multi-step commands. Memory is handled with vector-backed adaptive retrieval (ChromaDB), so agents remember context from earlier interactions and files. The extension system lets agents interact with the world — executing code, reading and writing files, searching the web and connecting to third-party APIs — while a web interface and REST API make it accessible to non-programmers and integrators alike. Everything is self-hosted with Docker.
AGiXT is completely free and open source under the MIT license. You self-host it and only pay for any third-party model APIs your agents use.
AGiXT is for developers, tinkerers and automation enthusiasts who want a self-hosted, provider-agnostic agent platform for building personal or business automations without being locked into one model vendor.