安装与连接
第三方代码:运行前请先阅读。HPSILab 审核过此说明,但不运营该服务器。
按客户端安装
claude mcp add --transport http jina-ai-mcp-server https://mcp.jina.ai/v1
由 HPSILab 根据本条目的端点和软件包,按各客户端官方文档的格式生成。参数、密钥和版本以官方说明为准。
官方说明
# From the official README: https://github.com/jina-ai/MCP
# Remote server (API key optional)
claude mcp add -s user --transport http jina https://mcp.jina.ai/v1 \
--header "Authorization: Bearer ${JINA_API_KEY}"远程 Endpoint
https://mcp.jina.ai/v1工具(12)
来自服务器公开端点的 tools/list,检查于 2026年11月22日。HPSILab 从不调用这些工具。
工具说明检查:未发现警告
按规则检查文字中是否有提示词注入或不安全请求的迹象,不是安全保证。
primerCurrent time, user location and network environment for the session. Use before answering anything time- or location-dependent.
guess_datetime_urlGuess when a page was published or last updated, with a confidence score. Checks HTTP headers, metadata, Schema.org, visible dates, feeds and sitemaps.
参数(1)
url· string · 必填
capture_screenshot_urlScreenshot a page as base64 JPEG. Use when the page must be seen rather than read.
参数(3)
url· string · 必填firstScreenOnly· booleanCapture only the first screen instead of the full page. Faster.return_url· booleanReturn URLs instead of base64 images.
read_urlFetch a URL (page or PDF) as clean markdown. Pass `question` to get only the passages answering it instead of the whole page - far cheaper than reading the full body into context.
参数(8)
url· 必填Page or PDF URL, or up to 5 URLs to read in parallel.withAllLinks· booleanAlso return every link on the page.withAllImages· booleanAlso return every image on the page.question· stringReturn only the passages answering this question, instead of the full body. Omit for the whole page.chunk_size· integerPassage size in words, default 100. Larger gives more surrounding context, smaller pinpoints the answer. Comparable across scripts. Needs `question`.topk· integerPassages to return, default 1. Needs `question`.ocr· booleanRead the rendered page as an image with jina-ocr-v1 instead of its HTML. Use for scanned documents and PDFs whose text is not in the markup, and to keep formulas and table structure. Parses ONE page per call, page 1 unless `page` says otherwise, and costs about 40x the tokens per page, so leave off unless the HTML path has failed you.page· integerWhich page the OCR pass should parse, 1-based. Needs `ocr`; the HTML path returns the whole document at once.
search_webSearch the web. Returns titles, URLs and search-engine snippets.
参数(6)
query· 必填num· integertbs· stringAge limit: qdr:h, qdr:d, qdr:w, qdr:m, qdr:y (hour to year).location· stringgl· stringCountry code, e.g. 'de'.hl· stringLanguage code, e.g. 'zh-cn'.
search_arxivSearch arXiv preprints (physics, maths, CS, quantitative biology and finance).
参数(3)
query· 必填num· integertbs· stringAge limit: qdr:h, qdr:d, qdr:w, qdr:m, qdr:y (hour to year).
search_ssrnSearch SSRN working papers (social science, economics, law, finance, management).
参数(3)
query· 必填num· integertbs· stringAge limit: qdr:h, qdr:d, qdr:w, qdr:m, qdr:y (hour to year).
search_jina_blogSearch Jina AI's news and blog at jina.ai/news: product announcements, model releases, technical write-ups.
参数(3)
query· 必填num· integertbs· stringAge limit: qdr:h, qdr:d, qdr:w, qdr:m, qdr:y (hour to year).
search_imagesSearch the web for images. Returns base64 JPEGs by default.
参数(7)
query· string · 必填num· integerreturn_url· booleanReturn URLs and metadata instead of base64 images.tbs· stringAge limit: qdr:h, qdr:d, qdr:w, qdr:m, qdr:y (hour to year).location· stringgl· stringCountry code, e.g. 'de'.hl· stringLanguage code, e.g. 'zh-cn'.
sort_by_relevanceRerank documents by relevance to a query, using Jina Reranker.
参数(3)
query· string · 必填documents· array · 必填Documents to rank, up to 1024.top_n· integer
deduplicate_stringsSelect the top-k semantically distinct strings from a list, dropping near-duplicates.
参数(2)
strings· array · 必填Strings to deduplicate, up to 1000.k· integerHow many to keep. Omit to pick k automatically from diminishing returns.
extract_pdfExtract figures, tables and equations from a PDF as images, by arXiv id or URL.
参数(4)
id· stringarXiv id, e.g. '2301.12345'. Give id or url.url· stringPDF URL on arxiv.org, ssrn.com, openreview.net or aclanthology.org; other hosts are refused. Give id or url.max_edge· numberLongest image edge in pixels, default 1024.type· stringComma-separated types to keep: figure, table, equation. Default all.
链接
更多「Agent 基础设施」类 MCP Server
- Hyperbrowser MCP Server
Scrape, crawl and extract structured data, or run browser agents, through Hyperbrowser's cloud browsers.
- Oxylabs MCP Server
Scrape and parse web pages, including JavaScript-heavy and geo-restricted sites, with Oxylabs.
- Bright Data MCP Server
Real-time web data for agents: search, scrape and browse sites without getting blocked, through Bright Data.
- Langfuse Prompt Management MCP
Fetch and use prompts managed in Langfuse, so agents and apps share versioned prompt templates.
- Docker MCP Gateway
Docker's gateway for running and managing MCP servers in isolated containers, with a shared catalog and secrets.
- Apify MCP Server
Run thousands of ready-made Apify Actors for web scraping, crawling and data extraction from your AI client.