reader

Why we included this project

Reader is the tool you reach for when a webpage won't cooperate with an LLM. Instead of pasting raw HTML and watching the model choke on nav bars and scripts, you put the service's prefix in front of the URL and get the page back as clean markdown, with headless browser rendering, PDF parsing, and captioned images handled for you. A companion search endpoint returns full page content for a query rather than just titles and snippets, so agent and RAG builders skip the crawling and extraction work they'd otherwise write by hand. The hosted API is free for production use, and the open-source branch runs locally via Docker when you want caching and data flow under your own control. For teams feeding web content into an LLM pipeline, that's a lot of plumbing you don't have to build.