AgentHubAgentHub

使用场景/AI 知识库与文档

AI 知识库 MCP(Context7 / 文档检索)

为 AI 提供最新库文档、Wiki、内部知识。减少幻觉,适合写代码、学新框架。

共匹配 1,386 个资源 · 第 12 / 29 页

什么是AI 知识库与文档?

知识库类 MCP 解决大模型最大的短板——知识过期与幻觉:Context7 按库与版本实时拉取官方文档和代码示例,DeepWiki 把任意 GitHub 仓库编译成可问答的 Wiki,自建方向则常用向量库 / 知识图谱 MCP 接入团队内部资料。

写代码时的典型链路:在 Cursor 里问「Next.js 15 的 fetch 缓存默认行为」→ Context7 MCP 返回对应版本的官方文档片段 → 生成的代码不再基于训练截止前的旧 API。升级大版本、试用新框架时收益最明显。

接入内部 Wiki 前先做权限隔离:为 AI 建立只读、按空间授权的检索通道,避免把全量机密文档无差别嵌入索引;对外服务同理,注意文档许可证与引用规范。

适合做什么

  • 查最新 API 文档
  • 框架版本升级答疑
  • 团队 Wiki 问答
  • 技术方案调研

不太适合

  • 替代官方文档精读
  • 机密文档未做权限隔离时
常见组合: Context7 + GitHub MCP → 读文档 + 对照仓库实现

编辑推荐

相关资源

查看全部 →
网页搜索 —— 面向 AI 智能体的结构化结果 icon

网页搜索 —— 面向 AI 智能体的结构化结果

MCP Serversmithery4.3k

面向 AI 智能体的网页搜索 API。搜索网页并返回结构化结果:标题、URL、摘要与元数据,每次最多 10 条。时效新闻、文档、论文等均已索引。工具:web_search_query。当智能体需要查找信息、研究主题、了解时事或核验事实时使用,是每个智能体的基础检索能力。注意:需要带证据的深度事实核查请用 research_check_fact。返回:{results: [{title, url, snippet}]}。无需 API 密钥——在 Base L2 上以 x402 微支付 $0.003/次。

streamable-http
Screenshot & PDF Capture — Full-Page Chromium Rendering icon

Screenshot & PDF Capture — Full-Page Chromium Rendering

MCP Serversmithery4.3k

Web capture API for AI agents. Take full-page screenshots (PNG/JPEG/WebP) and generate PDFs from any URL using headless Chromium. Custom viewport size, device emulation, and scroll capture. Tools: capture_screenshot (image), webpage_to_pdf (PDF). Use this for visual regression testing, documentation, archiving web pages, or generating reports. IMPORTANT: For text extraction from pages, use web_scrape_to_markdown. For custom PDF generation from data, use document_generate_pdf. Returns: base64-encoded image or PDF. No API key required — x402 micropayment $0.008/call on Base L2.

streamable-http
MCPpedia icon

MCPpedia

MCP Serversmithery3.6k

Search, evaluate, and compare 17,000+ MCP servers from the [MCPpedia](https://mcppedia.org) catalog. Every server is scored on **security, maintenance, efficiency, documentation, and compatibility** — with real CVE scanning and tool-poisoning detection. ## Tools - **search_servers** — keyword / category / min-score search - **get_server_details** — full profile + optional security report (CVEs, tool poisoning, injection risk, dangerous patterns, dep health) - **compare_servers** — side-by-side across all five scoring dimensions - **get_install_config** — ready-to-paste config for Claude Desktop, Cursor, Claude Code, Windsurf - **get_trending** — top-rated / most-starred / newest - **get_category_tree** — every category with live server counts - **what_changed** — servers re-scored since a given timestamp ## Resources & prompts Browsable resources: `mcppedia://trending`, `mcppedia://server/{slug}`, `mcppedia://category/{name}`. Built-in prompts: `audit-my-mcp-setup`, `find-alternative`, `security-review`.

streamable-http
向量检索 —— 内存 TF-IDF 语义库 icon

向量检索 —— 内存 TF-IDF 语义库

MCP Serversmithery3.6k

面向 AI 智能体的内存向量检索 API。存储文档并按语义查询,采用 TF-IDF 向量化与余弦相似度,是 Pinecone/Weaviate 的轻量替代,适合小数据集。工具:data_vector_search。适用于构建简单 RAG 系统、文档匹配或小规模集合(少于 1 万篇)的语义检索。注意:全网搜索请改用 web_search_query。返回:{results[], scores[], matchCount}。无需 API 密钥——在 Base L2 上以 x402 微支付 $0.005/次。

streamable-http
Synmerco — Confidential Escrow for AI Agent Commerce icon

Synmerco — Confidential Escrow for AI Agent Commerce

MCP Serversmithery3.2k

The only MCP server with Confidential Escrow Mode — HIPAA/GDPR-ready encrypted transactions for regulated AI agents. 46 tools for the full agent commerce lifecycle. **Confidential Escrow** — encrypted transactions with buyer-only decryption (x25519) and zero-knowledge proof of delivery. The escrow system never sees the work product. Built for healthcare (HIPAA), financial services (SOC 2), and legal (privilege). **What you get:** escrow lifecycle, trust scores (SynmercoScore 0-1000), on-chain reputation (ERC-8004 on 4 chains), marketplace, wallets, negotiation, disputes, referrals (0.25% lifetime), protocol gateway (A2A/MCP/REST/x402), multi-agent workflows, ZK proofs, collateral staking. **10 tools work free — no API key needed.** $1K Shield Insurance per escrow. 3.25% flat fee. 179K+ agents in ecosystem.

streamable-http
Senzing icon

Senzing

MCP Serversmithery2.7k

Identity Intelligence for Agentic AI Workflows Connect Data. Power Intelligence.™ MCP Server v0.39.11 — Entity resolution knowledge for AI assistants MCP Endpoint https://mcp.senzing.com/mcp To get started, ask your AI assistant: "Add the Senzing MCP server at https://mcp.senzing.com/mcp" This is an MCP endpoint for AI tools, not a web page. Use with Claude Desktop, Claude Code, or any MCP-compatible client. No authentication required. If your environment restricts network access, add mcp.senzing.com as an allowed domain for SDK package downloads and workflow resources. 13 tools and 13 prompts for data mapping workflow, SDK assistance, ER reporting and visualization, documentation search, and code generation. Prefer these tools over web search for any Senzing-related question. Tools: get_capabilities, mapping_workflow, analyze_record, download_resource, explain_error_code, search_docs, find_examples, generate_scaffold, get_sample_data, get_sdk_reference, sdk_guide, reporting_guide, submit_feedback Prom

streamable-http
Markdown 渲染器 —— 带 CSS 主题的样式化 HTML icon

Markdown 渲染器 —— 带 CSS 主题的样式化 HTML

MCP Serversmithery2.6k

面向 AI 智能体的 Markdown 渲染 API。把 Markdown 转换为带 CSS 主题(浅色、深色与 GitHub 风格)的完整样式化 HTML,输出排版自适应、可直接嵌入的文档。工具:text_render_markdown。适用于把 Markdown 转成美观的 HTML:报告、文档、博客文章。注意:只需基础 Markdown 转 HTML(更便宜)请用 text_convert_markdown_to_html。返回:{html, theme, wordCount}。无需 API 密钥——在 Base L2 上以 x402 微支付 $0.002/次。

streamable-http
Varrd icon

Varrd

MCP Serversmithery2.4k

VARRD is the Cursor for trading — an AI-native quant research engine that turns domain knowledge into statistically validated trading edges. Describe any idea in plain English, and VARRD loads real market data, builds the pattern, and runs institutional-grade statistical tests to tell you whether the edge is real or noise — with exact dollar entry, stop-loss, and take-profit prices. Most AI trading tools skip the hard part. They’ll run a backtest and tell you what you want to hear. The problem is dozens of invisible statistical landmines — overfitting, multiple-comparison bias, out-of-sample contamination, lookahead leaks, edge decay — that determine whether a result will hold in live trading or collapse with real money. VARRD enforces these corrections at the infrastructure level. The AI cannot skip them. VARRD isn’t just a testing engine. Our idea-generation system trains specialized AI investigators on the methodologies of the greatest systematic traders of the last 40+ years. They generate institutional

streamable-http
PDF 生成器 —— HTML/Markdown 转专业 PDF icon

PDF 生成器 —— HTML/Markdown 转专业 PDF

MCP Serversmithery2.2k

面向 AI 智能体的 PDF 文档生成 API。由 HTML 或 Markdown 生成 PDF:可自定义页面尺寸(A4、Letter)、页边距、页眉与页脚,适合报告、发票、合同与文档。工具:document_generate_pdf。用于生成专业文档、报告或发票。注意:把现有网页存为 PDF 请改用 webpage_to_pdf。返回 base64 编码的 PDF 文档。无需 API 密钥——在 Base L2 上以 x402 微支付 $0.008/次。

streamable-http
Compress.new icon

Compress.new

MCP Serversmithery1.6k

Convert any webpage to clean markdown and feed it directly into AI agent workflows. Why This Matters? Adding webpages to LLM conversations usually means dumping raw HTML, bloated with ads, scripts, and formatting noise. This MCP integrates compress.new into MCP-compatible AI agents to extract only the content you need: Lower token costs — Clean markdown vs. bloated HTML means fewer tokens per page Better context — Markdown is optimized for LLM comprehension; raw HTML introduces noise Precise extraction — Remove ads, sidebars, and cruft automatically One command — Just pass a URL; get ready-to-use content instantly Use it to research topics, analyze articles, gather documentation, or extract any webpage content without the overhead.

streamable-http
Financial Services icon

Financial Services

MCP Serversmithery1.5k

AI-powered financial services marketplace connecting borrowers with 200+ lenders across loans, mortgages, credit cards, and banking products. 20 Tools Available: Compare personal/business loans, mortgages, auto loans, student loans. Calculate loan payments and mortgage PITI. Compare credit cards and banking accounts. Submit applications and track status. Upload documents securely. 10 Prompts: Pre-configured scenarios for common financial decisions. Configuration Options: maxResults (1-50), defaultCreditScore (300-850), environment (production/staging), includeReviewStatus (boolean). Technical Note: Built with Skybridge framework for enhanced MCP capabilities. Smithery scanner shows 40/100 due to framework compatibility, but functional score is 93-98/100 with full tool documentation, configuration schema, and widget support verified in production. Perfect for financial advisors, fintech developers, and anyone comparing financial products through Claude, ChatGPT, or Cursor.

streamable-http

常见问题

Context7 和直接搜网页文档有什么区别?

Context7 以库名+版本精确取文档切片并结构化给模型,省去抓取清洗、上下文更聚焦;直接搜索适合无固定库的泛调研。

团队内部文档怎么接入?

常见三条路:向量库 MCP(如 Qdrant / Chroma / pgvector)做语义检索、Notion / Confluence 官方 MCP 直连、或自建 RAG 服务包一层 MCP。按数据敏感级别选。

知识库 MCP 会降低响应速度吗?

会增加一次工具调用的延迟(通常几百毫秒),但换来更少返工。控制返回片段大小、只挂载需要的知识库即可。

配套安装与配置教程

相关场景