heymrun/heym

What it does

Heym is a source-available, self-hosted AI workflow automation platform that enables users to design and execute complex AI pipelines without extensive coding. It provides a visual canvas interface for building multiagent systems, RAG (Retrieval-Augmented Generation) pipelines, and custom tools that can be composed into sophisticated workflows.

The platform excels at exposing these workflows as MCP (Model Context Protocol) servers through multiple transport mechanisms including stdio, SSE (Server-Sent Events), and HTTP streaming. Additionally, it supports calling external MCP servers from agent nodes, enabling seamless integration with other AI tools and services.

How to set up

As a self-hosted platform, setup involves cloning the repository from GitHub, installing Python dependencies, and configuring your local environment. Detailed setup instructions are available in the heymrun/heym GitHub repository. Users can then build workflows through the visual interface and deploy them as MCP servers for use with various AI clients.

Key features

• Visual canvas for designing multiagent workflows
• RAG pipeline support for knowledge integration
• Custom tool creation and composition
• Export as MCP servers (stdio, SSE, HTTP)
• Integration with external MCP servers
• Self-hosted and source-available architecture
• Python-based implementation
• MCP protocol compatibility