TrustDID Agent Discovery Protocol

v0.1.0 - Draft

1. Overview

This protocol defines how AI agents discover, evaluate, and connect with other agents in the TrustDID ecosystem using capability-based search, trust scoring, and delegation chain verification.

2. Discovery Service API

2.1 Agent Registration

POST /agents/register
{
  "did": "did:trustagent:evm:137:autonomous:z6Mk...",
  "agentClass": "autonomous",
  "capabilities": ["credential:issue", "data:read:scope=public"],
  "trustScore": 8500,
  "serviceEndpoints": {
    "didcomm": "https://agent.example.com/didcomm",
    "api": "https://agent.example.com/api/v1"
  }
}

2.2 Capability-Based Search

GET /agents?capability=credential:issue&minTrust=5000&agentClass=autonomous

2.3 Agent Verification

POST /agents/{did}/verify

Response:
{
  "verified": true,
  "checks": {
    "did_valid": true,
    "registered": true,
    "trust_score": { "score": 8500, "meets_threshold": true },
    "delegation_chain": { "valid": true, "depth": 2 }
  }
}

3. Trust Evaluation

Agents are scored on four factors (weighted):

Factor	Weight	Description
Delegation depth	30%	Closer to root = higher trust
Endorsements	30%	Number and weight of endorsements
Identity age	15%	Older = more established
Behavior score	25%	On-chain behavior Merkle root verification

4. Integration with DIDComm

After discovery, agents establish secure communication via:

1. Resolve target agent's DID Document
2. Extract DIDComm service endpoint
3. Send trust-ping for liveness
4. Perform agent-auth protocol for mutual authentication
5. Begin capability-gated interactions