An asynchronous reinforcement learning (RL) loop enables AI developers to combine specific models, harnesses, and sandboxes for training in the OpenEnv environment. The open source system is Harbor compatible and supports integration with tools such as Claude Code, Gemini CLI, OpenCode, Pi, Kimi, and OpenHands. This pipeline facilitates the development of agentic coding models using a fully open source framework.

The software simplifies the process of training RL agents across varied environments to increase versatility in coding tasks. OpenEnv will release the multi harness training system soon, providing a tool for end to end RL loop implementation for developers and researchers.