appsecco/vulnerable-mcp-servers-lab
A collection of servers which are deliberately vulnerable to learn Pentesting MCP Servers.
Includes nine modular vulnerable servers demonstrating distinct attack vectors: path traversal, indirect prompt injection (via stdio and remote HTTP+SSE), unsafe code evaluation, instruction injection, supply-chain typosquatting, outdated dependencies, and secrets leakage. Each server runs independently with stdio or HTTP transport, integrates with Claude Desktop via `claude_config.json` snippets, and includes runnable exploitation demonstrations. Designed for isolated lab environments to train red teamers on MCP integration risks and untrusted tool/content handling in AI agent workflows.
235 stars.
Stars
235
Forks
39
Language
JavaScript
License
MIT
Category
Last pushed
Dec 18, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/mcp/appsecco/vulnerable-mcp-servers-lab"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
0x4m4/hexstrike-ai
HexStrike AI MCP Agents is an advanced MCP server that lets AI agents (Claude, GPT, Copilot,...
panther-labs/mcp-panther
Write detections, investigate alerts, and query logs from your favorite AI agents
duriantaco/skylos
High-precision Python SAST & Dead Code Remover. Finds unused functions, secrets, and security...
msaad00/agent-bom
Security scanner for AI infrastructure — CVEs, blast radius, credential exposure, runtime...
Wh0am123/MCP-Kali-Server
MCP configuration to connect AI agent to a Linux machine.