
MCP Shamash
Enables security auditing, penetration testing, and compliance validation with tools like Semgrep, Trivy, Gitleaks, and OWASP ZAP. Features strict project boundary enforcement and supports OWASP, CIS, and NIST compliance frameworks.
README
MCP Shamash - Security Audit & Compliance Server
A Model Context Protocol (MCP) server for security auditing, penetration testing, and compliance validation with strict project boundary enforcement.
Features
- Project-Scoped Security Scanning: Never escapes project boundaries
- Multiple Security Tools: Semgrep, Trivy, Gitleaks, OWASP ZAP, and more
- Network Penetration Testing: Safe network scanning within project scope
- Compliance Validation: OWASP, CIS, NIST, ISO 27001 frameworks
- Containerized Execution: Isolated scanner execution with Docker
- Parallel Processing: Concurrent scanner execution for speed
- Intelligent Caching: Results cached for performance optimization
- Token Efficiency: <1000 tokens per operation with monitoring
- Comprehensive Audit Logging: Complete operation trails
- Real-Time Boundary Enforcement: Multi-layer security isolation
Quick Start
Installation
npm install
npm run build
Running the Server
npm start
Or for development:
npm run dev
Demo
Experience all features with the interactive demo:
node demo.js
This demonstrates:
- Real security tool integration (Semgrep, Trivy, Gitleaks, OWASP ZAP)
- Boundary enforcement (prevents external access)
- Network scanning within project scope
- Compliance validation (OWASP, CIS, NIST)
- Caching performance improvements
- Parallel scanner execution
MCP Integration
Configure in your MCP-compatible client:
{
"mcpServers": {
"shamash": {
"command": "node",
"args": ["/path/to/mcp_shamash/dist/index.js"]
}
}
}
Available Tools
scan_project
Comprehensive security scan of project directory.
{
"name": "scan_project",
"arguments": {
"path": "/path/to/project",
"profile": "standard",
"tools": ["semgrep", "trivy", "gitleaks"]
}
}
scan_network
Network scanning within project boundaries.
{
"name": "scan_network",
"arguments": {
"target": "127.0.0.1",
"ports": "80,443",
"serviceDetection": true
}
}
pentest_application
Penetration testing of deployed applications.
{
"name": "pentest_application",
"arguments": {
"targetUrl": "http://localhost:3000",
"testTypes": ["sql_injection", "xss", "csrf"],
"depth": "thorough"
}
}
check_compliance
Compliance framework validation.
{
"name": "check_compliance",
"arguments": {
"path": "/path/to/project",
"frameworks": ["OWASP", "CIS", "NIST"]
}
}
Security Boundaries
Project Scope Detection
- Automatic discovery of Docker Compose networks
- Kubernetes service detection
- Package.json analysis for Node.js apps
- Local service enumeration
Multi-Layer Enforcement
- Path Validation: Prevents directory traversal
- Network Boundaries: CIDR-based network restrictions
- Container Isolation: Docker security hardening
- Resource Limits: Memory, CPU, and process constraints
Blocked Operations
- System path access (
/etc
,/usr
,/var
) - External network scanning
- Management port access (22, 3389, 445)
- Privilege escalation attempts
Architecture
mcp-shamash/
├── src/
│ ├── core/ # MCP server core
│ ├── boundaries/ # Scope enforcement
│ ├── scanners/ # Tool integrations
│ ├── compliance/ # Framework validators
│ └── utils/ # Token management, audit logging
├── containers/ # Docker configurations
├── rules/ # Security rules
└── tests/ # Test suites
Development
Building
npm run build
Testing
npm test
npm run test:coverage
Linting
npm run lint
npm run format
Container Usage
Build Scanner Containers
# Build Semgrep scanner
docker build -f containers/Dockerfile.semgrep -t shamash-semgrep .
# Build all scanners
docker-compose -f containers/docker-compose.scanners.yml build
Run Isolated Scan
# Set target path and run scan
export SHAMASH_TARGET_PATH=/path/to/project
docker-compose -f containers/docker-compose.scanners.yml up semgrep
Configuration
Environment Variables
SHAMASH_MAX_TOKENS_PER_SCAN
: Token limit per scan (default: 1000)SHAMASH_MAX_TOKENS_PER_HOUR
: Hourly token limit (default: 50000)SHAMASH_AUDIT_LOG_PATH
: Audit log location (default: ./audit.log)
Project Configuration
Create .shamash.yml
in project root:
networks:
allowed:
- 172.20.0.0/16
- 127.0.0.1/32
blocked:
- 10.0.0.0/8
ports:
allowed: [80, 443, 3000, 8080]
blocked: [22, 3389, 445]
tools:
semgrep:
config: "auto"
timeout: 300
trivy:
severity: "HIGH,CRITICAL"
gitleaks:
entropy_threshold: 4.5
Compliance Frameworks
OWASP Top 10 Coverage
- A01: Broken Access Control
- A02: Cryptographic Failures
- A03: Injection
- A04: Insecure Design
- A05: Security Misconfiguration
- A06: Vulnerable Components
- A07: Authentication Failures
- A08: Software/Data Integrity
- A09: Security Logging
- A10: Server-Side Request Forgery
CIS Controls
- Inventory and Control of Assets
- Access Control Management
- Continuous Vulnerability Management
- Network Infrastructure Management
- Data Protection
NIST Cybersecurity Framework
- Identify: Asset management, governance
- Protect: Access control, data security
- Detect: Security monitoring, detection processes
- Respond: Response planning, incident management
- Recover: Recovery planning, improvements
Security Considerations
Defensive Only
- No offensive capabilities
- Read-only filesystem operations
- No credential harvesting
- Audit trail for all operations
Boundary Enforcement
- Multiple validation layers
- Real-time monitoring
- Automatic violation detection
- Emergency shutdown capability
Token Management
- Per-scan limits (1000 tokens)
- Rate limiting (5000/minute, 50000/hour)
- Usage tracking and reporting
License
MIT License
Contributing
- Fork the repository
- Create a feature branch
- Add tests for new functionality
- Ensure all tests pass
- Submit a pull request
Support
For issues and questions:
- Create an issue on GitHub
- Check the audit logs for troubleshooting
- Review boundary enforcement logs
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。
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。
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。

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。
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。
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。
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。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。

e2b-mcp-server
使用 MCP 通过 e2b 运行代码。