I upload paper.tex from ~/projects/anamnesis/. I get help from an AI, produce a revised version. It downloads to ~/Downloads/. Every time. Every tool. For thirty years. The browser knows where I uploaded from. The download could go back there. It doesn't. This seems obvious once you say it. Am I missing something? Is there a technical reason this can't work, or has nobody built it? A browser extension or desktop app could:

Capture upload origin path Store it for the session Default downloads to that path User overrides if needed

The conversation itself becomes provenance metadata - "this file went to Claude, here's what happened, here's what came back." Has anyone built this? If not, why not?