An officially maintained remote MCP server from Jina AI wrapping its Reader, Embeddings, and Reranker APIs into a broad toolset for web reading, search, deduplication, reranking, and PDF extraction, with server-side tool filtering to manage context usage and thorough documentation covering setup and troubleshooting across major MCP clients.
Strongest · Documentation 18/20 Weakest · Reliability 7/20
Static review · not runListed 2026-08-26
Read the FMRS scoring method →Runtime：Hosted (no local runtime) · Node.js
JINA_API_KEY
Jina AI API key (format jina_xxx), obtainable free at https://jina.ai; optional tools work without it under rate limits, but search and embeddings tools require it.
1) If your client supports remote MCP servers natively (Cursor, LM Studio, Claude Code), add url: https://mcp.jina.ai/v1 with an optional Authorization: Bearer ${JINA_API_KEY} header. 2) If your client lacks remote MCP support (e.g. some Claude Desktop setups), connect through the local mcp-remote proxy via npx, passing the server URL and Authorization header as args. 3) For Claude Code, run claude mcp add -s user --transport http jina https://mcp.jina.ai/v1 --header "Authorization: Bearer ${JINA_API_KEY}". 4) For OpenAI Codex, edit ~/.codex/config.toml to add an mcp_servers entry that launches mcp-remote. 5) Optionally filter which tools get registered server-side via query params, e.g. ?exclude_tags=parallel or ?include_tags=search,read, to save context window space.
{"mcpServers":{"jina-mcp-server":{"url":"https://mcp.jina.ai/v1","headers":{"Authorization":"Bearer ${JINA_API_KEY}"}}}}
Shown for Cursor. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.
{
"servers": {
"jina-mcp-server": {
"url": "https://mcp.jina.ai/v1",
"headers": {
"Authorization": "Bearer ${JINA_API_KEY}"
}
}
}
}
Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).
claude mcp add --transport http jina-mcp-server https://mcp.jina.ai/v1 --header 'Authorization: Bearer ${JINA_API_KEY}'
Run it in a terminal; replace any <…> placeholders with your own values first.
Check that tools like search_web and read_url appear in your MCP client's tool list; then ask a current-events question such as "What is the latest model from Jina AI?" — a fresh answer proves the connection works.
Once connected, you can ask your AI assistant things like:
No matching tools
Listed from the project's documentation, not tested by this site.
An officially maintained remote Model Context Protocol server from Jina AI that exposes 22 tools backed by the Reader, Embeddings, and Reranker APIs — covering URL-to-markdown extraction, web/image/arXiv/SSRN search, screenshotting, deduplication, classification, reranking, and PDF structure extraction. It is hosted at https://mcp.jina.ai/v1 using Streamable HTTP transport (the /sse path is kept only as a backward-compatible alias). Some tools work without an API key under rate limits; most require a Jina API key for full access. Server-side tool filtering via URL query parameters (include_tools/exclude_tools/include_tags/exclude_tags/max_tokens) lets clients register only the tools they need to save context tokens.
Manage your Hevy workout data from AI assistants via MCP.
Firecrawl's official MCP server for web search, scraping, and structured extraction for AI agents
One binary. One grammar. Evidence from the biomedical sources you already trust.
Encrypted, fully offline long-term memory for AI agents
Source revision 5d6eb191a75d Data synced 2026-10-12 Read the FMRS scoring method