Jina AI MCP
Jina AI Remote MCP Server
Jina AI's official remote MCP server: pages to Markdown, web search, arXiv, reranking and embeddings
- 897 ★ on the repository
- Last commit
- Apache-2.0
Extended access
Why this level
- Sends queries and page URLs to the external Jina AI service
- Uses a secret API key
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add mcp/jina-ai-remote-mcp-serverDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Sends queries and page URLs to the external Jina AI service
- Uses a secret API key
This is third-party code. Review the repository files before installing.
What it does
The server gives an agent tools built on Jina AI's Reader, Embeddings and Reranker APIs. It turns web pages into clean Markdown, captures screenshots, guesses publish dates, searches the web, images, arXiv and SSRN, returns BibTeX for papers and extracts tables and equations from PDFs. It has parallel search and read variants, document reranking, text classification and semantic deduplication. Tools can be filtered with URL parameters so unused descriptions do not take up model context.
Who it is for. For researchers, analysts and developers who need their agent to read and search the web.
Good fit when
- You need to read several pages as clean text
- You need fresh web results or arXiv papers
- You need to rank found documents by relevance
Not a fit when
- Data must not leave for an external service
- You need locally hosted search
Example request
Find recent arXiv papers on reranking, read the top three and summarize themMCP
- Transport
- http
- Authentication
- key is optional
Environment variables
JINA_API_KEYsecret- Jina API key sent in the Authorization header; required for search, reranking and embeddings
Limitations
Search, reranking and embeddings need a Jina API key; page reading works without one but with rate limits. The server is remote and all requests go to Jina AI. Claude Code, Claude Desktop and Cursor cap tool responses, so long pages are truncated with a note.
How to disable. In Claude Code run claude mcp remove -s user jina; in other clients delete the jina-mcp-server block from the MCP config.
FAQ
Does it work without a key?
Partly: page reading, screenshots and BibTeX search work without a key but are rate limited. Search needs a key.
How do I reduce the number of tools?
Add include_tags, exclude_tags, include_tools or exclude_tools to the endpoint URL.
Discussion
No questions or reviews yet.
Related
MCP serversOfficial Oxylabs MCP server: page scraping with JS rendering, Google and Amazon results, an AI crawler
A skill and the jina CLI: the agent turns any URL into clean Markdown and searches the web, including X posts and news sites
A local research MCP server: web, Reddit and YouTube search, maps and code exploration for a model in LM Studio or Cursor
Yandex and Google search MCP server
web-compass
An MCP web-search and content-extraction server via Yandex and Google on top of SerpAPI, geared toward RU/CIS results
MCP server for real-time web search: general and vertical search, batch queries and page content extraction
A unified search MCP server: one interface to Tavily, Brave, Kagi, Exa, Linkup, Firecrawl and GitHub search
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Jina AI MCP (this entry) | yes | key is optional | 897 | |
| qiaomu-markdown-proxy | no | 508 | a skill that reads a link and hands the agent the full text as | |
| jina-cli: web reading and search via Jina Reader | no | 311 | a skill and the jina CLI |
README in short
The README has a tool table noting which tools need a key. It gives configs for clients with remote server support, mcp-remote for the rest, and commands for Claude Code and Codex. It covers tool filtering via URL parameters and tag groups. Troubleshooting explains truncation of long responses, client-side tool caching and Windows issues.