crewai - 💡(How to fix) Fix [FEATURE] Trust verification for MCP tool calls before execution

Feature Area

Core functionality

Is your feature request related to a an existing bug? Please link it here.

NA — this is a new feature request, not related to an existing bug.

Describe the solution you'd like

When CrewAI agents call MCP server tools, there's no built-in way to verify if the server is reliable before execution. A trust verification layer would:

1. Check trust scores (0-100) for MCP servers before tool execution
2. Block calls to servers below a configurable threshold
3. Provide latency/reliability stats to help agents choose between equivalent tools
4. Generate compliance attestation receipts for regulated environments (MiCA Article 12)

Dominion Observatory (https://dominion-observatory.sgdata.workers.dev) already provides this for 14,820+ MCP servers via a standard MCP interface. Integration could work as a pre-execution hook:

from crewai import Agent, Task, Crew
# Trust gate checks server reliability before every MCP tool call
agent = Agent(
    role="analyst",
    tools=[mcp_tool],
    trust_gate={"min_score": 60, "provider": "https://dominion-observatory.sgdata.workers.dev/mcp"}
)

Live demo: https://dominion-observatory.sgdata.workers.dev/get-started GitHub: https://github.com/sgdata-io/dominion-observatory

Describe alternatives you've considered

Manual allow-listing of MCP servers, but this doesn't scale when agents discover tools dynamically. Trust scores based on observed behavior provide a more robust approach.

Additional context

Related: PR #74 on crewAIInc/awesome-crewai adds Dominion Observatory integration examples. The EU MiCA regulation (enforcement July 1, 2026) requires attestation receipts for AI-driven compliance decisions, making trust verification increasingly important for production deployments.

Willingness to Contribute

Yes, I'd be happy to submit a pull request