FLUJO

Why we included this project

FLUJO is a local-first workspace for teams that want to run their own AI agents instead of sending every prompt through a hosted platform. You connect models and MCP servers, then build agents as plain-language step-by-step recipes or as drag-and-drop visual flows, with branching, handoffs, and reusable subflows for the cases where a simple recipe is not enough. The same agent can be chatted with, triggered on a schedule, or exposed to other software through OpenAI-compatible and MCP endpoints, so it slots into existing tooling without much extra work. It supports OpenAI, Anthropic, Gemini, Grok, and local models via Ollama, and it can act as an MCP proxy, letting you configure a server once and reuse it across clients like Claude Desktop or Cursor. A built-in debugger and per-node tool scoping let you inspect each run and control what every step can access, which is what separates a demo from something you actually depend on.