heym

Heym describes itself as an AI-native workflow automation platform: you build, visualize and run workflows on a drag-and-drop canvas (Vue Flow, 30+ node types) without writing code, then deploy through Docker. Its stated difference is that AI is the execution model rather than a layer added to trigger-action automation. The stack is Python with FastAPI on the backend and Vue with TypeScript on the front end.

The capability list includes an LLM node and a full Agent node (tool calling, MCP connections, skills, Python tools, optional persistent memory), multi-agent orchestration where one agent calls named sub-agents and sub-workflows in parallel, human-in-the-loop pauses for approval, guardrails such as content filtering, built-in RAG with vector stores, an AI assistant that generates and wires nodes from a natural-language prompt, and an AI skill builder that previews the generated `SKILL.md`. The README also says execution history, global variables, observability and team controls ship in the free self-hostable product rather than behind an enterprise tier.

Start with `git clone`, copy `.env.example` to `.env` and run `./run.sh`. The license is important: Heym uses the Commons Clause on top of the MIT License, which the README itself calls source-available rather than OSI open source. You may use, modify and self-host it, but not sell it or offer it as a paid service; commercial licensing is available. It was last pushed on 2026-10-12. This review is based on the README and repository data; we have not run it ourselves.