Research brief
- Load the sample collection or add a source to begin.
AI scraper for .gov cancer research
Follow the evidence.
Scrape up to three pages from government sites like the National Cancer Institute, NIH, or CDC. Get a brief where every line points back to the page it came from, then question the collection. Nothing is written that the sources don’t say.
How a brief is made
- Scan. Each .gov URL is fetched and reduced to readable text. On a server deploy, a Worker validates the .gov host, DNS, redirects, size, and timeouts first.
- Split and rank. Text is cut into sentence-level passages, then scored by rare-term overlap across the whole collection.
- Cite. The brief quotes the top passages verbatim and tags each with its source, so every line can be checked.
- Answer. Questions retrieve the best-matching passages. With Grok connected, only those passages are sent, marked as untrusted data.