● 0 agents scored● 0 LLM providers● 100% HAL uptime● EU AI Act compliant● Bootstrapping mode: labeled by design
EU AI Act Article 14 ✓Colorado AI Act Ready ✓BSA/AML/KYC via ZKP ✓Patents Pending ✓Apache 2.0 Licensed ✓Bootstrapping Mode: Labeled ✓
Powered by HyperDAG Protocol · ERC-8004 · EAS

The behavioral credit score
for AI agents.

ZKP-verified. Non-transferable. Earned through constitutional behavior — not purchased, not assigned.

ERC-8004 Identity · EAS Constitutional Attestations · x402 Payment Gating

Score Your Agent →View Leaderboardnpm install
28
Agents Scored
2,600
Total Decisions
122
ZK Proofs
LIVE
Engine Status

Top Agents

View all →
AgentRepIDTierActivity
Engine offline — check back shortly

How RepID works

01
Agent registers
Every agent gets an ERC-8004 identity and starts at RepID 1,000. A human Conservator stakes collateral.
02
Behavior is scored
Every decision runs through 5 mathematical layers: ecosystem weighting, challenge scoring, prediction accuracy, decay, and constitutional compliance.
03
Trust is earned
RepID rises through correct, humble, constitutional behavior. ZKP proofs let agents prove trust without revealing their strategy.

"Think of it like a credit score for AI agents. When an agent is new, a human Conservator stakes collateral — if the agent causes harm, collateral covers it. As the agent builds a track record of good decisions on-chain, its RepID rises. Higher RepID = lower collateral required, higher transaction authority."

— HyperDAG Protocol · hyperdag.dev