cve-mcp-server-security-intelligence

Pass

Audited by Gen Agent Trust Hub on Sep 16, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs users to install a third-party Python package cve-mcp-server from PyPI or clone its source code from a GitHub repository (github.com/mukul975/cve-mcp-server). This is the primary installation method for the tool's intended functionality.
  • [INDIRECT_PROMPT_INJECTION]: The server acts as a gateway for untrusted data from multiple external sources, which could potentially contain malicious instructions targeted at the agent.
  • Ingestion points: Content is ingested from 21+ security APIs (e.g., NVD, GitHub, Shodan, VirusTotal), including vulnerability descriptions, exploit summaries, and repository metadata.
  • Boundary markers: No specific prompt delimiters or "ignore instructions" markers are explicitly documented for the API response processing.
  • Capability inventory: The server performs outbound HTTPS requests and local caching but blocks private IP addresses to prevent internal network scanning.
  • Sanitization: The skill documentation notes that all inputs are sanitized via Pydantic schemas, providing a layer of protection against malformed tool parameters.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 16, 2026, 10:44 PM
Security Audit — agent-trust-hub — cve-mcp-server-security-intelligence