dot-agent/nextpy

Implements compile-time prompt optimization and LLM session state management with KV cache reuse to reduce token overhead, plus structured output control that surpasses traditional prompt chaining. Includes syntax error detection for LLM-generated code, multiplatform agent deployment via .agent files, and optional Agentbox sandboxing for resource optimization. Built on FastAPI, React, and components from Guidance/DSPy/LangChain, with a generative UI layer using Reflex and Solara.

2,336 stars and 91 monthly downloads. No commits in the last 6 months. Available on PyPI.