Launching today
PageSift
Web content engine for AI agents: pages & PDFs → Markdown
0 followers
Web content engine for AI agents: pages & PDFs → Markdown
0 followers
PageSift turns URLs, PDFs and Office files into clean Markdown and structured data for AI agents. One self-hosted Rust binary: no per-page pricing, no external services. It scrapes, crawls, maps, batches and monitors sites, parses PDFs (including scanned, with built-in OCR), Word, Excel, PowerPoint, EPUB and CSV, and offers CLI, Firecrawl-compatible API, Jina-compatible endpoints, MCP server and Python/Node SDKs. Fast (PDF 2.4 ms/page; 3,500–4,400 pages/s in ~300 MB) and open source (AGPL-3.0).
Free
Launch Team
CloudflareYour next app belongs on Cloudflare Workers. Deploy for free
Promoted
Hi everyone — I'm Zhou, the solo developer of PageSift. I built it because the crawler feeding my research platform was too slow and too dirty, so I wrote my own in Rust. Fifteen days after the first public release, 1.0 closes the feature list: OCR, change monitoring, browser actions, self-healing extraction, and an agent endpoint. Everything is measured, and the benchmark methodology (including where PageSift loses) is public in the repo. Happy to answer anything — and head-to-head benchmarks against your current stack are very welcome.