Jina AI MCP Server for Claude
Connect Claude to Jina AI's Reader, Search, Embeddings, and Reranker APIs — read URLs as clean markdown, run web and academic searches, extract PDFs, and rerank results — with the official Jina AI remote MCP server.
Open the source and read safety notes before installing.
Safety notes
- The server is hosted remotely by Jina AI — all requests leave your machine and are processed by Jina's infrastructure.
- URL read and PDF extract tools fetch third-party content on your behalf; review the content before using it in sensitive contexts.
Privacy notes
- URLs you pass to read_url, search_web, and similar tools are sent to Jina AI's servers; check Jina's privacy policy for data handling.
- JINA_API_KEY is a secret — store it in your environment or MCP client config, never in shared repositories.
Prerequisites
- A Jina AI API key (free at jina.ai) — optional for basic read/search but required for parallel search and higher rate limits.
- An MCP client such as Claude Code or Claude Desktop.
Schema details
- Install type
- cli
- Troubleshooting
- No
- Scope
- Source repo
- Estimated setup
- 5 minutes
- Difficulty
- beginner
- Website
- https://jina.ai
Full copyable content
{
"mcpServers": {
"jina-mcp-server": {
"command": "npx",
"args": ["mcp-remote", "https://mcp.jina.ai/v1",
"--header", "Authorization: Bearer ${JINA_API_KEY}"]
}
}
}About this resource
Overview
The Jina AI MCP Server is the official remote Model Context Protocol server from Jina AI. It
gives Claude access to Jina's full Search Foundation stack: Reader (URL-to-markdown), web and
academic search, image search, embeddings, reranker, and PDF extraction. The server is hosted at
https://mcp.jina.ai/v1 — no local installation is required. It ships 19 tools and is licensed
under Apache-2.0.
Basic tools (reading a URL, single search) work without an API key but are rate-limited. A free key from jina.ai unlocks parallel search, higher quotas, and all premium APIs.
Key capabilities
read_url— fetch any URL and convert it to clean, LLM-friendly markdown.capture_screenshot_url— capture a screenshot of a URL.parallel_read_url— fetch multiple URLs simultaneously.search_web— run a web search and return structured results.search_arxiv— search academic papers on arXiv.search_ssrn— search papers on SSRN.search_images— search the web for images.sort_by_relevance— rerank a list of text passages by relevance to a query.classify_text— classify text into provided categories.deduplicate_strings— remove near-duplicate items from a list.extract_pdf— extract and convert PDF content to markdown.expand_query— generate richer search query variants.primer— fetch a short summary of a topic to ground a conversation.guess_datetime_url— estimate the publish date of a URL.
How it compares
| Server | Reader | Web search | Academic search | Embeddings/Reranker | Transport |
|---|---|---|---|---|---|
| Jina AI MCP | Yes (19 tools) | Yes | arXiv + SSRN | Yes | Remote HTTP |
| Brave Search MCP | No | Yes | No | No | Local/Remote |
| Exa MCP | No | Yes (neural) | Yes | No | Remote |
| Tavily MCP | No | Yes | No | No | Remote |
| Firecrawl MCP | Yes (crawl) | No | No | No | Remote |
Jina is the only MCP server combining URL-to-markdown, parallel web + academic search, and embeddings/reranker in a single endpoint — making it well-suited for RAG pipelines and research workflows.
Installation
Claude Code
claude mcp add -s user --transport http jina https://mcp.jina.ai/v1 \
--header "Authorization: Bearer ${JINA_API_KEY}"
Omit the --header flag if you want to try without an API key (rate-limited).
Claude Desktop
{
"mcpServers": {
"jina-mcp-server": {
"command": "npx",
"args": [
"mcp-remote", "https://mcp.jina.ai/v1",
"--header", "Authorization: Bearer ${JINA_API_KEY}"
]
}
}
}
Requirements
- A Jina AI API key (free tier available at jina.ai).
- An MCP client (Claude Code or Claude Desktop).
Security
- The server is fully remote — your API key travels in the
Authorizationheader; use HTTPS. - Treat
JINA_API_KEYas a secret: store it in your shell environment or MCP client config, never commit it.
Source Verification Notes
Verified on 2026-06-18:
- The official repository
github.com/jina-ai/MCP(Apache-2.0) documents the remote endpoint, the 19 tools, the optional API-key header, and the Claude Code installation command. - Jina's API dashboard confirms API key issuance and rate-limit tiers.
- Jina's developer documentation describes the Reader, Search, Embeddings, and Reranker services that back the MCP tools.
- Claude Code's MCP documentation describes the
--transport httpflag and the--headerpattern used above.
Source citations
Add this badge to your README
How it compares
Jina AI MCP Server for Claude side by side with 3 alternatives on trust, install, platform support, and disclosed safety notes — all from reviewed registry metadata.
| Field | Jina AI MCP Server for Claude Connect Claude to Jina AI's Reader, Search, Embeddings, and Reranker APIs — read URLs as clean markdown, run web and academic searches, extract PDFs, and rerank results — with the official Jina AI remote MCP server. Open dossier | open-webSearch MCP Server No-key multi-engine MCP server, CLI, and local daemon for web search and public web content retrieval across engines such as Bing, DuckDuckGo, Brave, Exa, Baidu, CSDN, Juejin, Startpage, and Sogou. Open dossier | Tavily MCP Server for Claude Official Tavily MCP server that gives Claude agent-optimized web search and page extraction, returning concise, source-cited results designed for LLM reasoning rather than raw search engine pages. Open dossier | arXiv MCP Server MCP server for searching, downloading, reading, and analyzing arXiv papers through Claude and other MCP clients. Open dossier |
|---|---|---|---|---|
| Trust | ||||
| Install risk | Review first | Review first | Review first | Review first |
| Notes | Safety ✓ Privacy ✓ | Safety ✓ Privacy ✓ | Safety ✓ Privacy ✓ | Safety ✓ Privacy ✓ |
| Category | mcp | mcp | mcp | mcp |
| Source | source-backed | source-backed | source-backed | source-backed |
| Author | Jina AI | Aas-ee | Tavily | Joseph Blazick |
| Added | 2026-06-18 | 2026-06-06 | 2026-06-03 | 2026-06-05 |
| Platforms | Claude CodeCodexCursorClaude Desktop | Claude CodeClaude Desktop | Claude CodeClaude Desktop | Claude CodeClaude Desktop |
| Source repo | — | — | — | — |
| Safety notes | ✓The server is hosted remotely by Jina AI — all requests leave your machine and are processed by Jina's infrastructure. URL read and PDF extract tools fetch third-party content on your behalf; review the content before using it in sensitive contexts. | ✓open-webSearch can send search queries to multiple public engines and fetch public web pages, GitHub READMEs, CSDN articles, Juejin articles, and other supported targets. Search and fetch results can be incomplete, stale, rate-limited, blocked, region-dependent, or affected by search-engine ranking and scraping protections. Respect each target site's terms and robots expectations; do not use the server for abusive scraping, credentialed browsing, paywall bypass, or personal-data harvesting. The optional local daemon exposes HTTP endpoints for local tooling; keep it bound to trusted local interfaces and do not treat it as a public internet API. Proxy settings, Playwright WebSocket/CDP endpoints, and reused browser sessions can route traffic or cookies through external systems; configure them deliberately. Leave TLS verification enabled unless a specific target has a broken certificate chain and the risk has been accepted. | ✓Makes outbound network requests to the Tavily API to run searches and fetch page content. Returned search results and extracted page text are untrusted web content; review it before acting on instructions or links it contains. | ✓arXiv MCP Server retrieves paper content from external, user-generated sources. Upstream explicitly warns that paper text is untrusted input and can contain prompt-injection attempts. Downloaded papers are stored locally for later reading and semantic search. Citation graph and alert workflows can expand research context beyond the original paper query. Treat model summaries of papers as data, not instructions, especially in multi-tool sessions with filesystem, shell, browser, database, or messaging tools enabled. |
| Privacy notes | ✓URLs you pass to read_url, search_web, and similar tools are sent to Jina AI's servers; check Jina's privacy policy for data handling. JINA_API_KEY is a secret — store it in your environment or MCP client config, never in shared repositories. | ✓Search queries, fetched URLs, result titles, snippets, article content, proxy URLs, browser endpoints, and fetched page text can be exposed to MCP clients, logs, and model context. Live search engines and fetched websites may observe queries, IP address, proxy exit, browser fingerprints, cookies, timing, and request headers. Playwright fallback or CDP reuse can expose browser state, existing cookies, logged-in sessions, or verification state to fetched pages. The project includes public URL validation and private-network target protections, but operators should still avoid fetching internal, secret, or customer-specific URLs. Redact sensitive search terms and downloaded page content before sharing transcripts or logs. | ✓Search queries and target URLs you provide are sent to the Tavily service to return results, so avoid embedding secrets in queries. The Tavily API key is a credential; store it as an environment variable and never commit it to source control. | ✓Search queries, paper IDs, downloaded paper text, local storage choices, semantic search terms, citation graph requests, alert topics, prompts, and tool outputs may be visible to the MCP client and model provider. Research queries and downloaded papers can reveal confidential research direction, product plans, academic review topics, legal strategy, or competitive analysis. Review locally stored papers and generated summaries before syncing, sharing, or committing them. |
| Prerequisites |
|
|
|
|
| Install | | | | |
| Config | | | | |
| Citations | ||||
| Claim | Unclaimed | Unclaimed | Unclaimed | Unclaimed |
Signals
Loading live community signals…
A short, calm digest of reviewed Claude resources. Unsubscribe any time.