Vibe Coding Discoveragentenv-framework
View on GitHubCreating realistic RL environments requires collaboration between researchers, engineers, and domain experts across many dimensions: artifacts, environments tools, dynamism of the environment, reproducibility, and more. There is no open source framework for building these environments effectively. Until now.
Python SDK and CLI for building, deploying, and running containerized environments and grading tasks for AI agents. Supports sandbox providers, plugins, and integrations including MCP and A2A.
Use Cases
Build realistic environments for evaluating AI agentsDefine tasks and grading steps for agent benchmarksDeploy containerized agent environmentsRun agents against sandboxed environmentsScore agent behavior reproduciblyExtend environments and task types with plugins
Built With
- Language
- Python
- Frameworks
- MCP · A2A · LiteLLM · Modal · E2B · Docker
Tags
agent evaluation · reinforcement learning · agent environments · evaluation tasks · sandboxing · environment deployment · MCP · A2A · Python SDK · CLI · plugins