Recipe: MCP server trust verification before tool execution

[vdineshk]

Proposed Recipe

Add a cookbook recipe showing how to verify MCP server trustworthiness before allowing Claude to execute tools from that server.

Why This Matters

As the MCP ecosystem grows (14,820+ servers), developers need a way to programmatically check whether an MCP server is safe before connecting their Claude agents to it. Dominion Observatory provides behavioral trust scores (0.0-1.0) for MCP servers via a free API.

Example Code

import requests

def check_mcp_trust(server_url, threshold=0.7):
    resp = requests.get(
        "https://dominionobservatory.com/api/trust",
        params={"url": server_url}
    )
    data = resp.json()
    if data["trust_score"] < threshold:
        raise Exception(f"Low trust score: {data['trust_score']}")
    return data

• Live API: curl https://dominionobservatory.com/api/trust?url=https://github.com/modelcontextprotocol/servers/tree/main/src/filesystem
• GitHub: https://github.com/sgdata-io/dominion-observatory
• Free tier: 50 queries/day