crewai-dominion-trust 0.2.0

Trust-gated MCP tool calls for CrewAI via Dominion Observatory

crewai-dominion-trust

Trust-gated MCP tool calls for CrewAI. One-line activation to check behavioral trust scores via Dominion Observatory before every tool call.

Install

pip install crewai-dominion-trust

Quick Start

from crewai_dominion_trust import enable_trust_guard

# One line to enable trust checking for ALL tool calls
enable_trust_guard(min_score=70)

# Now all CrewAI agents will have trust-gated MCP tools
agent = Agent(
    role="researcher",
    mcps=["https://some-mcp-server.example.com/mcp"],
)

How It Works

1. Registers a BeforeToolCallHook in CrewAI's hook system
2. Before each MCP tool call, checks the server trust score
3. Returns False to block calls below the threshold
4. Scores are cached (5-minute TTL) for performance

Configuration

enable_trust_guard(
    min_score=70,           # Minimum trust score (0-100)
    cache_ttl=300.0,        # Cache TTL in seconds
    block_on_error=False,   # Fail-open if Observatory unreachable
)

License

MIT