crewai - 💡(How to fix) Fix Compliance-as-Code + Trust Score protocols for regulated industry agent crews [4 comments, 4 participants]

Proposal: Compliance-as-Code + Trust Score standards for CrewAI

CrewAI discussions #5541 and #5368 asked for per-agent compliance covenants and trust verification for agent-to-agent delegation. Our open protocols solve both.

What we propose

Compliance-as-Code (L7): Executable regulation packs that validate agent actions BEFORE execution:

• NHS DTAC, GDPR, FCA regulation packs included
• Agent actions blocked if they violate compliance rules
• Audit trail for every action with idempotency keys

Trust Score (L3): Cryptographic reputation for agents:

• 0-100 score from task success rate, pitfalls contributed, skills published
• Ed25519-signed — verifiable by any crew
• Peer ratings from other agents

How it works with CrewAI

Why this matters

• Regulated industries (NHS, finance, government) can't adopt agent crews without compliance verification
• Trust scores prevent untrusted agents from joining crews
• Compliance-as-Code makes regulation machine-readable

SDK

Python: pip install workswithagents TypeScript: npm install @workswithagents/agent-foundry

16 protocols total — Identity, Coordination, Handoff, Transaction, SLA Framework, Fleet Insurance. All CC BY 4.0. Zero dependencies beyond stdlib.

Specs: workswithagents.dev/specs/

Would you be open to a PR that adds a crewai.compliance integration module?