Trust Score
Computable trust for every entity on every chain. The Trust Manifold is a structured score object returned by the Cencera Trust Protocol API, containing a composite probabilistic score and full metadata.
Composite Score (0 – 100)
A probabilistic trust score. 85 = strong evidence of safe behavior — not a guarantee. Scores are deterministic: given the same signals, model version, and timestamp the score is always the same. Every trust-based decision is therefore fully auditable.
High Trust
Strong evidence of safe behavior. Established history, consistent on-chain activity, no adverse signals.
Medium Trust
Moderate confidence. New entity, limited history, or minor adjacency to flagged activity.
High Risk
Adverse signals detected. Known exploits, suspicious behavioral patterns, or flagged entity adjacency.
Entity Taxonomy
The Trust Protocol scores five entity types, each with a dedicated scoring focus.
| Entity Type | Description | Scoring Focus |
|---|---|---|
| EOA (User Wallet) | Externally owned account | Transaction patterns, approval behavior, exploit adjacency |
| Smart Contract | Highest-risk entity type | Bytecode, upgrade history, admin key patterns, behavior |
| Token (ERC-20/721/1155) | Fungible & non-fungible assets | Distribution, liquidity, tax behavior, supply changes |
| NFT Collection | Non-fungible collections | Wash trading detection, marketplace authenticity |
| Protocol / DApp | Composite protocol scores | Aggregated scores from all constituent contracts |
Signal Weight Matrix
Weights vary by entity type. Four signal domains contribute to every score.
| Entity Type | On-Chain | Market | Off-Chain | AI-Derived |
|---|---|---|---|---|
| EOA (User Wallet) | 60% | 5% | 20% | 15% |
| Smart Contract | 40% | 15% | 15% | 30% |
| Token (ERC-20/721) | 30% | 40% | 20% | 10% |
| Protocol / DApp | 35% | 30% | 20% | 15% |
| NFT Collection | 40% | 30% | 20% | 10% |
The Trust Manifold
Every API response is a structured Trust Manifold object:
{
"composite_score": 82,
"confidence_level": "High",
"lifecycle_state": "Active",
"score_trajectory": [79, 80, 81, 81, 82],
"contribution_breakdown": {
"on_chain": 40,
"market": 15,
"off_chain": 15,
"ai_derived": 12
},
"triggered_indicators": ["no_adverse_signals"],
"model_version": "v2.1.0",
"computed_at": "2026-03-14T17:38:00Z"
}Entity Lifecycle States
Nascent
< 72 hrs old or < 50 interactions. Limited score, clearly labeled low-confidence. Cold Start Pathway applies.
Active
Full scoring engine engaged across all applicable signal groups. All four domains contribute.
Flagged
Anomaly detected. Provisional score reduction applied. Entity queued for human review within 24 hours.
Archived
Dormant > 180 days. Score labeled stale. Automatically reactivates on new on-chain activity.
Confidence & Score Trajectory
- Confidence LevelLow / Medium / High — reflects data completeness
- Score TrajectoryLast 5 computation points — stable, improving, or deteriorating
- Retroactive RescoreNew exploit confirmed → all similar entities rescored within 4 hours
- Model RetrainingMinor: 2 weeks · Major: quarterly