OWASP Agentic Security MCP

OWASP agentic AI security — prompt injection detection, tool poisoning prevention, agent trust boundaries by MEOK AI Labs

GitHub

Owasp Agentic MCP

MCP server for owasp agentic mcp operations

PyPI License: MIT MEOK AI Labs

Overview

Owasp Agentic MCP provides AI-powered tools via the Model Context Protocol (MCP).

Tools

ToolDescription
assess_agent_securityFull OWASP Agentic AI Top 10 security assessment.
check_prompt_injectionCheck text for prompt injection attack patterns.
check_tool_poisoningCheck a tool for name/description manipulation (tool poisoning).
check_excessive_agencyAssess agent for excessive permissions (least privilege).
check_data_leakageAssess cross-context data exposure risks.

Installation

pip install meok-owasp-agentic-mcp

Usage with Claude Desktop

Add to your Claude Desktop MCP config (claude_desktop_config.json):

{
  "mcpServers": {
    "owasp-agentic-mcp": {
      "command": "python",
      "args": ["-m", "meok_owasp_agentic_mcp.server"]
    }
  }
}

Usage with FastMCP

from mcp.server.fastmcp import FastMCP

# This server exposes 5 tool(s) via MCP
# See server.py for full implementation

License

MIT © MEOK AI Labs

Máy chủ liên quan