@opensyber/mcp

MCP (Model Context Protocol) server for OpenSyber. Gives AI agents like Claude Desktop, Cursor, and Claude Code direct access to security monitoring, dependency scanning, and threat intelligence.

Installation

npm install @opensyber/mcp
# or
pnpm add @opensyber/mcp

Configuration

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

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

Cursor

Add to .cursor/mcp.json in your project root:

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

Claude Code

claude mcp add opensyber -- npx -y @opensyber/mcp

Available Tools

opensyber_scan_dependency

Check if an npm package is safe to install. Returns a safety verdict (safe / suspicious / malicious) with reasons and known CVEs.

Parameters:

• package (required) — npm package name (e.g. "lodash")
• version (optional) — specific version to check (e.g. "4.17.21")

Example prompt: "Is the event-stream npm package safe to use?"

opensyber_check_security

Get a security score for a project. Returns a breakdown across 8 categories: dependencies, secrets, authentication, input validation, encryption, logging, access control, and infrastructure.

Parameters:

• projectPath (optional) — absolute path to the project root

Example prompt: "Run a security check on my project."

opensyber_query_threats

Get current AI agent threat intelligence. Returns recent threats including prompt injection, supply chain attacks, and model exfiltration attempts.

Parameters:

• severity (optional) — minimum severity: "critical", "high", or "medium"

Example prompt: "What are the latest critical threats targeting AI agents?"

opensyber_list_skills

Browse the OpenSyber audited skill marketplace. Lists available security skills that agents can install.

Parameters:

• category (optional) — filter by "monitoring", "scanning", "compliance", "networking", or "infrastructure"

Example prompt: "Show me available security scanning skills."

opensyber_protect

Generate a security configuration for a web framework. Returns integration code with TokenForge device binding, rate limiting, CORS, and CSP headers.

Parameters:

• framework (required) — "express", "hono", "nextjs", or "fastify"

Example prompt: "Generate security config for my Hono API."

Development

# Run locally
pnpm dev

# Build
pnpm build

# Test with MCP Inspector
npx @modelcontextprotocol/inspector dist/index.js

License

MIT