agentica

Why we included this project

Teams already building TypeScript backends get a direct path into AI agents here: you register functions from a TypeScript class, a Swagger/OpenAPI document, or an MCP server, and Agentica runs the function-calling loop against your LLM of choice. You skip orchestration graphs and prompt glue entirely. The compiler-driven core builds the calling schema from your actual types, and its validation feedback catches malformed AI arguments, so you are not debugging tool calls by hand. Growing the agent means adding or removing functions, which keeps it aligned with your existing service surface rather than a generic assistant. For a Node/TypeScript shop picking an agent framework, that is a strong practical fit.