Jina MCP Server

A Model Context Protocol server providing access to Jina AI's suite of web tools including Reader, Embeddings, and Reranker APIs.

Features

• URL Reading & Screenshots: Extract clean markdown content from web pages and capture high-quality screenshots
• Multi-Source Search: Search the web, arXiv papers, SSRN publications, images, and Jina AI blog
• Academic Tools: Search BibTeX citations, extract PDF content including figures and tables from arXiv papers
• Content Analysis: Rerank documents by relevance, classify text, deduplicate strings/images semantically
• Parallel Operations: Execute multiple searches or reads concurrently for comprehensive coverage
• Query Enhancement: Expand and rewrite search queries, analyze webpage timestamps
• Contextual Awareness: Get localized, time-aware context with the primer tool

Tools

Reading & Extraction

• read_url - Extract clean, structured content from web pages as markdown
• parallel_read_url - Read multiple web pages in parallel
• capture_screenshot_url - Capture high-quality screenshots of web pages
• extract_pdf - Extract figures, tables, and equations from PDF documents using layout detection
• guess_datetime_url - Analyze web pages for last update/publish datetime with confidence scores

Search

• search_web - Search the entire web for current information and news
• search_arxiv - Search academic papers and preprints on arXiv
• search_ssrn - Search academic papers on SSRN (Social Science Research Network)
• search_images - Search for images across the web (similar to Google Images)
• search_jina_blog - Search Jina AI news and blog posts
• search_bibtex - Search for academic papers and return BibTeX citations (DBLP + Semantic Scholar)
• parallel_search_web - Run multiple web searches in parallel
• parallel_search_arxiv - Run multiple arXiv searches in parallel
• parallel_search_ssrn - Run multiple SSRN searches in parallel

AI-Powered Analysis

• sort_by_relevance - Rerank documents by relevance to a query using Jina Reranker API
• classify_text - Classify texts into user-defined labels using Jina Embeddings API
• deduplicate_strings - Get top-k semantically unique strings via submodular optimization
• deduplicate_images - Get top-k semantically unique images via submodular optimization

Utilities

• primer - Get current contextual information for localized, time-aware responses
• expand_query - Expand and rewrite search queries based on query expansion model

Tool Filtering

Every MCP tool requires the LLM to pre-allocate tokens in its context window. For LLMs with limited context windows, you can filter tools server-side via query parameters:

Query Parameters

• exclude_tools - Comma-separated tool names to exclude
• include_tools - Comma-separated tool names to include
• exclude_tags - Comma-separated tags to exclude (search, parallel, read, utility, rerank)
• include_tags - Comma-separated tags to include

Examples

Exclude parallel tools: https://mcp.jina.ai/v1?exclude_tags=parallel

Only include search and read tools: https://mcp.jina.ai/v1?include_tags=search,read

API Key

Some tools work without an API key but have rate limits. For higher rate limits and better performance, get a free Jina API key from https://jina.ai.

Tools requiring API key: search_web, search_arxiv, search_ssrn, search_images, expand_query, parallel_search_, sort_by_relevance, classify_text, deduplicate_, extract_pdf

Tools optional API key: read_url, capture_screenshot_url, parallel_read_url (work without key but with rate limits)

Tools not requiring API key: primer, guess_datetime_url, search_jina_blog, search_bibtex