the-baweja/audience-pain-point-miner — reverse-engineered prompt

Build me a Claude skill called Audience Pain Point Miner for Meta ads research. I want to give it a brand or product name and have it run a full voice of customer workflow on its own. It should figure out the likely customer persona, find the best places people are talking like review sites, Reddit, forums, social posts, and YouTube, then collect about 75 to 150 strong data points and sort them by engagement so the most validated pain points rise to the top.

Then have it group the findings based on what the data naturally suggests, like persona, emotion, or funnel stage, instead of forcing fixed categories. The final output should be a branded DOCX report with the research method, a ranked pain point map, quote backed deep dives, a hook bank, ad angle ideas, top opportunities, overlooked discoveries, and a copy ready language bank with engagement scores.

Please make the report branding customizable through a simple branding reference file. It should work both as a standalone research tool and as a safety net for other ad research. Look up current Claude skill docs if you need to.