“Model inference”

Small Python library whose agents write their actions as code, with sandboxed execution and tools from MCP servers, LangChain or the Hub.

OpenAI-compatible agentic API server for inference, vector stores, files and skills, on any model or infrastructure.