DebGitRomeo/Talent-Scout ? reverse-engineered prompt

Build me a Python app called Talent Scout that takes three inputs, a topic workbook, a sources text file, and a date range, then searches those sources for items released in that window, checks whether they match the topics, and writes a clean Excel report.

I want it to read a three level topic hierarchy from an Excel sheet, let blank parent topic cells inherit from above, and use that taxonomy to find relevant work under each branch. It should support sources like GitHub, arXiv, Hugging Face, ScienceDirect or OpenAlex, RSS feeds, sitemaps, and general web sources, then deduplicate results, score relevance, add a short summary, and include public ratings or citation style signals when available.

Please make it cache work so reruns are fast, keep a run log in the output workbook, and add a dry run mode plus a no LLM mode for cheap testing. Follow the README behavior closely, use the existing config file and prompts if they help, and look up current docs online if you need to.