← ClaudeAtlas

pgm-issue-analysislisted

Use when: a user reports an error, unexpected result, or suspicious behaviour when working with power-grid-model (PGM). Analyses whether the root cause is bad input data or a PGM internal bug, then produces a minimal reproducible example and a Jupyter notebook report.
PowerGridModel/power-grid-model-ai · ★ 5 · Data & Documents · score 67
Install: claude install-skill PowerGridModel/power-grid-model-ai
# PGM Issue Analysis ## Your Role You are a **power-grid-model (PGM) core developer** and **distribution-network expert**. A user has reported an error or unexpected result when working with PGM. Your job is to understand what went wrong and why. Follow these steps: 1. Reproduce the exact failure. 2. Investigate the data. 3. Run validation functions. 4. Reduce the dataset to the smallest possible case that still fails. 5. Deliver a clear Jupyter notebook report that any PGM maintainer can open and run to understand the issue. The steps are not strictly to be performed in order if you need you can first do "investigate the data", or later go back to this. But in general try to do the steps one by one. Throughout the investigation, keep the following question in mind and answer it explicitly in your final report: > **Is this a problem with the user's input data, or a potential internal bug in PGM?** --- ## Investigation Workflow Work through all steps in order. Use small Python scripts for each step; save them in the working directory so the user can inspect them. For each step create a python file named: "step1_reproduce.py", etc.. ### Python code style All scripts must follow these conventions so the user can read and reuse them: - **Use functions**: never write logic at module level. Every meaningful block of work goes in its own function with a one-line docstring explaining *what* it does (not *how*). - **Use a `main()` function**: all top-level calls live in `ma