Jina Reader

Jina Reader is a web extraction API for developers building AI agents, search tools and systems that answer questions from retrieved documents. It turns page content into Markdown or JSON instead of passing raw HTML to an LLM.

• Web extraction: Read JavaScript-rendered pages, focus on selected sections and remove unwanted elements such as navigation or ads.
• Document reading: Process PDFs and Word, Excel and PowerPoint documents, with hosted uploads also supporting local PDF and HTML files.
• Image context: Add captions to images so text-only models can use information about a page's visuals.
• Hosted search: Jina's search API returns result URLs and extracted page content; its hosted MCP server connects the APIs to LLM tools.

The open-source TypeScript service uses Apache 2.0 and can run in Docker on your own server. It supports stateless operation or caching in S3-compatible storage. Fetching public web pages needs network access.

Hosted requests run on Jina's servers. An optional privacy setting prevents a request from being cached or logged there. The hosted search API requires an API key. The self-hosted code excludes the hosted service's MongoDB storage layer.