Proposal: Integrate AgentSign trust scoring with ANS agent records

[razashariff]

Summary

ANS handles agent discovery (DNS for agents). AgentSign handles agent trust verification. These are naturally complementary.

The Gap

ANS provides X.509 certificates for agent identity and a registry for discovery. However, once an agent is discovered, the requesting party has no way to assess:

• Has this agent been security-tested?
• What is its trust level?
• Has it passed a structured onboarding pipeline?
• Is its code attested (hash verified)?

What AgentSign Adds

AgentSign provides a trust scoring and behavioral verification layer:

• Trust scores (0-100) based on 5 factors: code attestation, execution verification rate, success rate, history depth, pipeline stage
• Pipeline stages: INTAKE -> VETTING -> TESTING -> DEV_APPROVED -> PROD_APPROVED -> ACTIVE
• 13-point SDLC scanner aligned to OWASP Agentic Top 10
• Cryptographic passports: Self-contained signed JSON with full agent metadata
• MCP Verification Gate: Identity + trust verification before tool access

Proposed Integration

1. Extend ANS agent records with optional trust_score and pipeline_stage fields
2. AgentSign as trust provider: ANS queries AgentSign for trust verification during agent resolution
3. Combined flow: ANS resolves agent endpoint -> AgentSign verifies trust -> tool access granted/denied
4. Passport in ANS metadata: Include AgentSign passport URL in ANS agent records for offline verification

Links

• Website: https://agentsign.dev
• GitHub SDK: https://github.com/razashariff/agentsign-sdk (MIT)
• npm: https://www.npmjs.com/package/agentsign
• Patent: GB2604808.2 (pending, UKIPO)

Happy to discuss or submit a PR implementing this integration.