Drift can read PDFs (datasheets, spec docs, papers) and web pages (vendor pages, GitHub READMEs, tutorials) directly. Point it at a file or URL and it’ll extract the parts relevant to your prompt — dimensions, part numbers, code snippets, protocol details — without you having to copy-paste.
PDFs
Attach a PDF path (local or URL) the same way you’d reference a file, and ask a question:
Drift decides internally whether to answer from a focused read (fast, cheap) or a full read (thorough). You don’t have to pick.
Password-protected PDFs are supported. If the PDF is encrypted, Drift decrypts it server-side before reading. No extra flags needed.
Web pages
Give Drift a URL — Drift fetches the page, reads it, and answers:
If you don’t know the exact URL, ask Drift to search:
What’s blocked
For safety, Drift refuses to fetch:
- Private IP ranges (
10.*, 192.168.*, 172.16-31.*, 127.*) and other internal-network targets
file:// and other non-http(s) schemes
Public URLs on the internet work; anything that would target your local network or filesystem won’t. 
