tribe

Why we included this project

Most multi-agent frameworks force you to write the orchestration graph by hand, but Tribe flips that: you assemble your team of LLM agents in a drag-and-drop visual workspace and get running in minutes. Underneath it builds on LangGraph, so you can arrange agents in sequential pipelines or as a hierarchical team where a leader delegates subtasks, and when a built-in skill isn't enough you can still drop into Python code. Teams keep persistent conversation state, can call external tools behind human-approval gates, answer from a knowledge base through RAG, and run local open-weight models via Ollama. There's a Docker-based deployment with a public API and multi-tenancy baked in, so a small team can stand up a real service rather than a demo. If you'd rather spend your time on the agents themselves than on wiring the coordination layer, this is a good place to start.