Skip to main content
Lucid’s reputation system builds trust scores for AI assets from real traffic data. Unlike self-reported metrics, Lucid reputation is backed by cryptographic receipts — every score is traceable to verified interactions.

​
Architecture

Reputation operates across three layers:

​
How Scores Are Computed

Reputation scores combine multiple signals:
  1. Feedback — Users submit scores (1-100) per interaction, linked to a receipt hash
  2. Validation — Third parties validate receipts, confirming interactions occurred
  3. Revocation — Submitters can revoke feedback; the score adjusts atomically
  4. Traffic volume — More verified interactions increase confidence
  5. Payment history — Consistent payments improve credit tier
The on-chain program maintains rolling statistics per passport: feedback count, validation count, total score, and average score (stored with 2-decimal precision).

​
Submitting Feedback

Feedback requires a valid receipt hash, ensuring that only users who actually interacted with the asset can rate it.

​
Reputation Tiers (PayReputation)

At the gateway level, agents accumulate a credit tier based on payment history: Higher tiers unlock benefits like reduced payment verification overhead and higher trust in multi-agent delegation.

​
Querying Reputation

​
Cross-Network Sync

Lucid reputation syncs with external identity standards: Lucid extends Metaplex identity beyond agents to all AI assets (models, compute, tools, datasets), creating a bidirectional reputation mesh.