CVE Search MCP Server
Enables searching and analyzing CVEs and vulnerabilities from multiple sources, optimized for PR review scenarios to help developers identify the latest security issues.
README
CVE Search MCP Server
A Model Context Protocol (MCP) server for CVE and vulnerability searching, optimized for PR review scenarios. Helps developers and security teams identify the latest CVEs and vulnerabilities — including those that postdate an LLM's training data.
Features
- 8 tools covering CVE lookup, bulk scanning, keyword search, product search, recent CVEs, high-severity alerts, database stats, and detailed CVSS breakdowns
- Multi-source: NVD, GitHub Advisory, OSV, CIRCL — searched concurrently and deduplicated
- Smart normalization: "Node.js", "Spring Boot", "log4j2" all resolve correctly
- Optional auth:
GITHUB_TOKENandNVD_API_KEYfor higher rate limits - 3 transports: stdio (default), SSE, Streamable HTTP
Installation
Prerequisites
- Python 3.10+
- uv package manager
Install
cd cve-search
uv sync
Development Setup
uv sync --extra dev
uv run --extra dev pytest
uv run black src/
uv run ruff check src/
Configuration
| Env Var | Default | Description |
|---|---|---|
GITHUB_TOKEN |
none | GitHub personal access token. Raises GitHub Advisory API rate limit from 60 to 5000 req/hr. |
NVD_API_KEY |
none | NVD API key. Raises rate limit from 5 to 50 req/30s. Get one at nvd.nist.gov. |
CVE_SEARCH_TIMEOUT |
20 |
HTTP request timeout in seconds. |
CVE_SEARCH_MAX_RESULTS |
100 |
Maximum results returned per tool call. |
Running
# stdio (default — for Claude Desktop/IDE)
uv run python main.py
# SSE
uv run python main.py --transport sse --host 127.0.0.1 --port 8000
# Streamable HTTP (MCP spec 2025-06-18+)
uv run python main.py --transport streamable-http --host 127.0.0.1 --port 8000
Claude Desktop Config
{
"mcpServers": {
"cve-search": {
"command": "uv",
"args": ["--directory", "/path/to/cve-search", "run", "python", "main.py"],
"env": {
"GITHUB_TOKEN": "your-token-here",
"NVD_API_KEY": "your-key-here"
}
}
}
}
Tools
| Tool | Description | Speed |
|---|---|---|
search_cve_by_id |
Look up a specific CVE by ID (e.g. CVE-2021-44228) | Fast |
bulk_cve_lookup |
Look up up to 20 CVE IDs in one call — ideal for scanning PR dependency lists | Fast |
search_vulnerabilities_by_product |
Search by vendor/product name (e.g. vendor="apache", product="struts") | Slow (10-15s) |
get_recent_cves |
Get CVEs from the last N days | Fast |
check_high_severity_cves |
Get CVSS ≥ 7.0 CVEs from the last N days | Fast |
search_by_keyword |
Smart multi-source keyword search (NVD + GitHub Advisory + OSV) | Fast |
get_vulnerability_stats |
Database stats: total CVE count, last updated timestamp | Fast |
cvss_score_lookup |
Detailed CVSS v3/v4 breakdown for a CVE (base score, vector string, per-metric) | Fast |
PR Review Workflow
Scan a list of CVE IDs from a dependency audit
bulk_cve_lookup(["CVE-2021-44228", "CVE-2023-44487", "CVE-2024-12345"])
Search for vulnerabilities in a technology being introduced
search_by_keyword("spring boot")
Check high-severity CVEs published this week
check_high_severity_cves(7)
Get detailed CVSS breakdown for a flagged CVE
cvss_score_lookup("CVE-2021-44228")
Project Structure
cve-search/
├── src/mcp_server_cve_search/
│ ├── config.py # Config from env vars
│ ├── server.py # FastMCP app + transport dispatch
│ ├── tools/ # One module per tool group
│ │ ├── cve_lookup.py # search_cve_by_id, bulk_cve_lookup
│ │ ├── product_search.py
│ │ ├── recent_cves.py # get_recent_cves, check_high_severity_cves
│ │ ├── keyword_search.py
│ │ ├── stats.py # get_vulnerability_stats
│ │ └── cvss.py # cvss_score_lookup
│ ├── sources/ # One client per API
│ │ ├── circl.py # CIRCL CVE Search
│ │ ├── nvd.py # NVD/NIST (optional API key)
│ │ ├── github.py # GitHub Advisory (optional token)
│ │ └── osv.py # OSV (Google)
│ └── utils/
│ ├── severity.py # CVSS score helpers
│ ├── normalization.py # Keyword normalization + tech mappings
│ └── formatting.py # Summary/alert formatting
├── tests/
├── examples/
├── main.py
├── test_server.py # Manual live-API integration test
└── pyproject.toml
Data Sources
| Source | URL | Notes |
|---|---|---|
| CIRCL CVE Search | cve.circl.lu | Primary source; no auth required |
| NVD (NIST) | nvd.nist.gov | Richest CVSS data; optional API key |
| GitHub Advisory | github.com/advisories | Optional token for higher rate limits |
| OSV (Google) | osv.dev | Open source vulnerability database |
License
MIT License
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。