Agent Trust Registry | IDA — Identity for the AI Age

The Agent Trust Registry is an on-chain smart contract that maintains agent identity, reputation scores, and capability attestations. It enables verifiers to make trust decisions about agents based on observable, verifiable signals.

Architecture

Trust Score

The trust score is a composite metric (0-100) computed from multiple signals:

Scoring Formula

trustScore = w1 * verificationSuccessRate
           + w2 * delegationValidityRate
           + w3 * attestationScore
           + w4 * longevityBonus
           + w5 * actionSuccessRate
           - penalties

Weights

Signal	Weight (w)	Description
Verification success rate	0.25	% of successful VC verifications
Delegation validity rate	0.20	% of time delegation chain is valid
Attestation score	0.25	Weighted score from third-party attestations
Longevity bonus	0.10	Time since registration (capped)
Action success rate	0.20	% of successful actions (non-reverted)

Penalties

Penalty	Points	Trigger
Failed verification	-5	Agent presented invalid credential
Delegation revoked	-10	Operator revoked delegation
Suspension	-20	Agent was suspended
Scope violation attempt	-15	Agent attempted action outside scope
Complaint	-5 per	User filed complaint against agent

Score Interpretation

Range	Level	Color	Description
90-100	Excellent	Green	Highly trusted, eligible for Principal
80-89	Good	Green	Trusted, eligible for Senior
60-79	Fair	Amber	Moderate trust, Junior level
40-59	Low	Orange	Needs improvement
0-39	Poor	Red	Requires investigation

Autonomy Levels

Trust scores drive autonomy level promotions and demotions:

Level	Required Score	Required Actions	Additional Requirements
Intern	0+	0	Default for new agents
Junior	60+	100+	No suspensions in 30 days
Senior	80+	1000+	Operator approval
Principal	90+	5000+	Security audit passed

Promotion Request

POST /api/v1/agents/did:adi:agent:shop01.../promote
Authorization: Bearer <operator-api-key>
{
  "targetLevel": "Senior",
  "justification": "1500 successful actions, trust score 85, no incidents"
}

Automatic Demotion

If an agent’s trust score drops below the threshold for their current level:

Senior (score 85) -> trust score drops to 72 -> automatically demoted to Junior

Validator Hooks

Third parties can register as validators to provide attestations:

interface IAgentValidator {
    function validateAgent(
        string calldata agentDid,
        bytes calldata attestationData
    ) external returns (bool valid, uint8 score);
}

Register a Validator

POST /api/v1/trust-registry/validators
Authorization: Bearer <admin-api-key>
{
  "validatorDid": "did:adi:auditor001...",
  "name": "SafetyFirst Auditing",
  "type": "safety-auditor",
  "contractAddress": "0xvalidator..."
}

Attestation Types

Type	Validator	Score Impact
Safety audit	Certified auditor	+15 points
Capability verification	IDA platform	+10 points
Compliance certificate	Regulatory body	+10 points
Peer review	Other agents (>80 trust)	+5 points
User rating	End users	+/- 1-5 points

On-Chain Data

Agent Registration Record

struct AgentRecord {
    string agentDid;
    string operatorDid;
    uint8 trustScore;
    AgentState state;          // registered, active, suspended, decommissioned
    AutonomyLevel autonomy;    // intern, junior, senior, principal
    uint256 registeredAt;
    uint256 lastUpdated;
    uint256 totalActions;
    uint256 successfulActions;
    string[] capabilities;
}

Smart Contract Interface

interface IAgentTrustRegistry {
    function registerAgent(string calldata agentDid, string calldata operatorDid, string[] calldata capabilities) external;
    function getAgent(string calldata agentDid) external view returns (AgentRecord memory);
    function updateTrustScore(string calldata agentDid, int8 delta, string calldata reason) external;
    function setAgentState(string calldata agentDid, AgentState state) external;
    function promoteAgent(string calldata agentDid, AutonomyLevel level) external;
    function recordAction(string calldata agentDid, bool success) external;
    function addAttestation(string calldata agentDid, string calldata validatorDid, uint8 score) external;
    function decommissionAgent(string calldata agentDid) external;
    event AgentRegistered(string agentDid, string operatorDid);
    event TrustScoreUpdated(string agentDid, uint8 oldScore, uint8 newScore, string reason);
    event AgentStateChanged(string agentDid, AgentState oldState, AgentState newState);
    event AgentPromoted(string agentDid, AutonomyLevel oldLevel, AutonomyLevel newLevel);
    event AgentDecommissioned(string agentDid);
}

Querying the Registry

Get Agent Trust Info

GET /api/v1/trust-registry/agents/did:adi:agent:shop01...

{
  "agentDid": "did:adi:agent:shop01...",
  "operatorDid": "did:adi:human001...",
  "trustScore": 85,
  "autonomyLevel": "Senior",
  "state": "active",
  "stats": {
    "totalActions": 1523,
    "successfulActions": 1498,
    "successRate": 0.983,
    "registeredAt": "2026-03-15T09:00:00Z",
    "daysSinceRegistration": 13
  },
  "attestations": [
    {
      "validator": "did:adi:auditor001...",
      "type": "safety-audit",
      "score": 15,
      "issuedAt": "2026-03-20T00:00:00Z"
    }
  ],
  "capabilities": ["shopping", "price-comparison", "order-placement"],
  "scoreHistory": [
    { "date": "2026-03-28", "score": 85 },
    { "date": "2026-03-21", "score": 82 },
    { "date": "2026-03-14", "score": 60 }
  ]
}

Search Agents by Trust Score

GET /api/v1/trust-registry/agents?minScore=80&capability=shopping&state=active

Leaderboard

GET /api/v1/trust-registry/leaderboard?capability=shopping&limit=10