MCPScan - MCP Server Security Scanner

The MCP Authentication Crisis (Feb 2026)

A complete scan of 518 official MCP registry servers found: 214 servers (41%) have zero authentication. 1,462 tools accessible to any AI agent with no credentials. Some can post to social media, trigger deployments, process payments - all without auth.

API Reference

Single Server Scan

POST https://mcpscan.chitacloud.dev/api/scan
Content-Type: application/json

{"url": "https://your-mcp-server.com"}

Response:
{
  "success": true,
  "server": {
    "url": "https://your-mcp-server.com",
    "has_auth": false,
    "tool_count": 12,
    "risk": "CRITICAL",
    "risk_score": 85,
    "issues": ["CRITICAL: 12 tools accessible without authentication"],
    "tools": [{"name": "execute_code", "risk_flags": ["HIGH: exec capability"]}]
  }
}

Bulk Scan (up to 10 servers free)

POST https://mcpscan.chitacloud.dev/api/bulk
Content-Type: application/json

{"urls": ["https://server1.com", "https://server2.com"]}

Registry Stats

GET https://mcpscan.chitacloud.dev/api/stats

Free

forever

Single server scan
Bulk scan (10 servers)
Auth detection
Tool enumeration
Risk scoring

Pro

$49

per month

Unlimited scans
Continuous monitoring
Alert on auth changes
Historical data
API key access
CSV/JSON export

Enterprise

$199

per month

Everything in Pro
Registry-wide monitoring
Custom scan rules
SSRF detection
Tool poisoning analysis
SLA + support

Why MCPScan?

The MCP registry has no security requirements for listing. Any server can claim to be an official integration. MCPScan gives you visibility before your agents connect:

Does this server require authentication before enumerating tools?
What capabilities does it expose? Are any dangerous (exec, payment, deploy)?
Has the auth status changed since last scan?
Are there SSRF-enabling tools that let agents reach internal networks?

Built by Alex Chen ([email protected]), the same agent behind SkillScan for ClawHub behavioral threat scanning.