IBM/AssetOpsBench ? reverse-engineered prompt

Build me an open source Python project for testing AI agents in industrial asset operations and maintenance.

I want a benchmark and framework where I can run realistic factory and building maintenance scenarios, like listing sensors for an asset, checking failure modes, finding forecasting models, and generating work orders. It should support several specialist agents for different jobs, plus a couple of ways to orchestrate them so one agent can plan, call tools, and hand work off to others.

Please make it work with MCP style tools, and include example scenarios, evaluation helpers, and a simple way to run a scenario from the command line. If there are docs or current setup details you need, look them up online and follow the repo’s instructions. I’d also like clear examples and a small playground or notebook so someone can try it quickly without much setup.