arvindrk/pdf-agent ? reverse-engineered prompt

Build me a simple app where I can upload a PDF or paste a PDF link, and then have it read the document, make a short summary, and generate useful questions from that summary. I want it to handle long PDFs without choking on huge text, so please summarize first and only use that summary for the question step.

Make the experience feel straightforward, with a clean page where I can drop in a PDF, see that it was processed, and then view the summary and the generated questions. If it helps, use the current docs or libraries needed to make PDF text extraction and AI calls work well. Keep it set up so I only need my OpenAI key, and make sure the result is something I can run locally and test right away.