NVD漏洞数据库

NVD — National Vulnerability Database

NIST's National Vulnerability Database. Every CVE (Common Vulnerabilities and Exposures) ever published — software vulnerabilities, severity scores (CVSS), affected products (CPEs), references. The authoritative source for "is this software version vulnerable?" Free, no auth (light rate limit; key recommended).

Part of Pipeworx — an MCP gateway connecting AI agents to 1856+ live data sources. This is an independent, unofficial integration — not affiliated with, endorsed by, or published by the upstream provider.

Why this matters for AI agents

For security analysis, supply-chain risk assessment, or "what CVEs affect dependency X?" the NVD is the source. Where commercial vulnerability databases add curation, the NVD is the raw federal record. Pair with USPTO patents for security IP, SEC EDGAR for breach disclosures.

Common flows:

• CVE lookup. Find specific CVE by ID for full record.
• Search by product / version. "What CVEs affect Apache Log4j 2.x?" → keyword + CPE filter.
• Recent CVEs by severity. Critical and high-severity disclosures published recently.
• CVSS scoring. Each CVE has CVSS v2, v3.0, and v3.1 scores; agents should use v3.x for current analysis.

Auth

NVD's REST API is free; an unauthenticated client gets ~5 requests per 30s. Get a free API key at https://nvd.nist.gov/developers/request-an-api-key for ~50 requests per 30s. Pass via _apiKey.

Severity classes (CVSS v3)

For agent triage, "High and Critical, last 90 days" is the common attention slice.

Common pitfalls

• CPE matching is fiddly. CPE (Common Platform Enumeration) is the controlled vocabulary for "this CVE affects this product version." Software names in CPE often differ from how marketing names them. Use NVD's CPE search to find the right CPE before searching CVEs.
• CVE coverage isn't complete. Some bugs are quietly patched without CVE assignment. Conversely, not every CVE is exploitable in practice. Triage by environment.
• Severity scoring is not exploitability. A Critical CVSS score on a feature you don't use is irrelevant; a Medium CVSS on something exposed to the internet is worse than the score implies. Pair with EPSS (Exploit Prediction Scoring System) when available.
• Reserved vs published. Reserved CVEs ("RESERVED") are placeholders awaiting public disclosure. The actual content lives in description once published. Filter vulnStatus for what's actually known.
• References lag. Patches and exploit-detection signatures often appear before the NVD record updates. For real-time vulnerability response, layer GitHub Security Advisories or vendor channels on top.
• Modified vs published date. The "modified" date often reflects re-scoring or reference updates, not new findings. For "what was disclosed this week," sort by publishedDate.

Quick Start

Add to your MCP client (Claude Desktop, Cursor, Windsurf, etc.):

{
  "mcpServers": {
    "nvd": {
      "url": "https://gateway.pipeworx.io/nvd/mcp"
    }
  }
}

What this endpoint actually serves

tools/list at https://gateway.pipeworx.io/nvd/mcp returns the tools in the table above plus the shared Pipeworx meta-tools — ask_pipeworx, discover_tools, search_within, remember/recall and the rest of the gateway-wide set. So the tool count you see is larger than this table: a single-pack endpoint currently lists roughly 30 shared tools alongside the pack's own. The connection's initialize response states its exact scope, and is the authoritative answer for a given day.

This is deliberate, not multiplexing by accident. The meta-tools are what let a scoped connection answer a question this pack does not cover — via ask_pipeworx, which routes across the whole catalog — without you adding a second MCP server. There is currently no way to mount a pack endpoint without them; if the extra schemas cost you more context than the routing is worth, connect to the full gateway once rather than to several pack endpoints.

Or connect to the full Pipeworx gateway to get every pack's tools listed directly, instead of just this one's:

{
  "mcpServers": {
    "pipeworx": {
      "url": "https://gateway.pipeworx.io/mcp"
    }
  }
}

Both URLs reach the same gateway and the same 1856+ data sources. The only difference is which pack's tools are listed directly; ask_pipeworx reaches all of them from either one.

No MCP client? Call it over HTTP

curl -X POST https://gateway.pipeworx.io/v1/tools/search_cves \
  -H 'Content-Type: application/json' \
  -d '{"query":"log4j"}'

No account needed for the first calls. Inspect any tool: GET https://gateway.pipeworx.io/v1/tools/search_cves. Find one: POST https://gateway.pipeworx.io/v1/tools/search_packs with {"query":"..."}.

Run it from npm

@pipeworx/mcp-nvd is a local stdio bridge to the Pipeworx gateway: your MCP client starts it with npx, and it forwards every call to the same https://gateway.pipeworx.io/nvd/mcp endpoint shown above. It needs network access; there is no offline mode.

{
  "mcpServers": {
    "nvd": {
      "command": "npx",
      "args": ["-y", "@pipeworx/mcp-nvd"]
    }
  }
}

Without a key it calls anonymously, capped at 50 calls a day. A free account raises that to 200 a day: get a key at https://pipeworx.io/signup and add "env": { "PIPEWORX_API_KEY": "<your key>" } to the server entry. The key is sent as a Bearer token, the same way a hosted connection sends it, so calls are metered and billed exactly as they are there.

It lists only this pack's tools by default. Set PIPEWORX_META_TOOLS=1 to also list ask_pipeworx and the other gateway-wide tools.

Behind an HTTPS proxy: Node's built-in fetch does not read HTTPS_PROXY by default. Newer Node releases honour it when NODE_USE_ENV_PROXY=1 is set; check your Node version's documentation.

Using with ask_pipeworx

Instead of calling tools directly, you can ask questions in plain English — this works on the pack endpoint above as well as on the full gateway:

ask_pipeworx({ question: "your question about Nvd data" })

The gateway picks the right tool and fills the arguments automatically.

More

• Docs and guides
• pipeworx.io

License

MIT