ToolTrust Scanner

Scan MCP servers for prompt injection, data exfiltration, and privilege escalation before your AI agent trusts them.

Urgent security update ToolTrust detects and blocks known compromised MCP-related package versions, including the LiteLLM / TeamPCP supply-chain exploit.

Live UI

• Browse the public registry: https://www.tooltrust.dev/
• Review findings in the browser before trusting or installing a server
• Compare grades across popular MCP servers

Use with Claude Code / Cursor / Claude Desktop

Add ToolTrust as an MCP server:

{
  "mcpServers": {
    "tooltrust": {
      "command": "npx",
      "args": ["-y", "tooltrust-mcp"]
    }
  }
}

Then ask your agent to:

• tooltrust_scan_config to scan all configured MCP servers
• tooltrust_scan_server to scan one specific MCP server
• Full MCP tool list: Usage Guide

What It Catches

• Prompt injection and tool poisoning hidden in descriptions
• Excessive permissions such as exec, network, db, and fs
• Supply-chain CVEs and known compromised package versions
• Privilege escalation and arbitrary code execution patterns
• Typosquatting, tool shadowing, and insecure secret handling
• Missing rate-limit, timeout, or retry configuration on risky tools

Full rule catalog: Developer Guide · tooltrust.dev

More Ways to Use ToolTrust

• Usage Guide
• Developer Guide
• Security Docs

Links

• GitHub: AgentSafe-AI/tooltrust-scanner
• Live UI: tooltrust.dev