model-compose

What it does\nmodel-compose is a lightweight alternative to LangChain that uses a Docker-compose inspired approach to define AI systems. It allows developers to describe models, agents, tools, and workflows in a single YAML file, ensuring portability across local, containerized, and cloud environments.\n\n## Tools\n- Workflow Orchestrator: Chains jobs with conditional logic and parallel execution.\n- Agent Component: Builds autonomous agents with tool use and planning capabilities.\n- HTTP/WebSocket Client: Connects to any external API or LLM provider.\n- Vector-Store Integration: Native support for Chroma, FAISS, Milvus, and Qdrant.\n- Protocol Adapter: Instantly serve AI systems over HTTP REST, WebSocket, or MCP.\n\n## Installation\nTo use as an MCP server, configure your claude_desktop_config.json to point to the model-compose runtime (after running model-compose up):\n\njson\n{\n \"mcpServers\": {\n \"model-compose\": {\n \"command\": \"node\",\n \"args\": [\"path/to/sse-bridge.mjs\", \"http://localhost:8080/sse\"]\n }\n }\n}\n\n\n## Supported hosts\n- Claude Desktop