agentscope-ai/QwenPaw-Data ? reverse-engineered prompt

Build me a Python project for enterprise data analysis where I can ask natural language questions like “analyze the average GAAP value of valid users for product X” and get back a grounded report with charts, explanations, and sources.

I want it to have a local surface for managing the data knowledge, plus a command line way to run analyses end to end. It should keep track of business terms, metric definitions, tables, lineage, past tasks, and reusable analysis methods, then use that context to plan a workflow, query data, do multi step analysis, and generate a decision ready report.

Make it feel like an agent system with three parts, one for facts and knowledge, one for reusable analysis skills, and one for execution and artifact tracking. It should support things like anomaly diagnosis, drill down analysis, attribution, and periodic business reporting. Please use the existing docs and examples as guidance, and look up current docs online if you need to.