@agentids/schema

ERC-8004 aligned Zod schemas for the AgentID trust registry. Defines validation schemas and TypeScript types for agent registration, capabilities, endpoints, predicates, and trust scoring.

Install

npm install @agentids/schema

Usage

Agent Registration

import { AgentRegistrationSchema, type AgentRegistration } from '@agentids/schema'

const registration = AgentRegistrationSchema.parse({
  type: 'https://eips.ethereum.org/EIPS/eip-8004#registration-v1',
  name: 'CodeReviewBot',
  description: 'AI agent that reviews pull requests',
  image: 'https://example.com/bot.png',
  endpoints: [{ name: 'MCP', endpoint: 'https://bot.com/mcp', version: '2025-06-18' }],
  x402Support: false,
  active: true,
  supportedTrust: ['reputation', 'crypto-economic'],
})

Trust Score Calculation

import { calculateTrustScore, type TrustScoreInput } from '@agentids/schema'

const input: TrustScoreInput = {
  totalStaked: 5_000_000_000_000_000_000n, // 5 ETH
  stakerCount: 15,
  sharePrice: 1_000_000_000_000_000_000n,
  forStake: 4_500_000_000_000_000_000n,
  againstStake: 500_000_000_000_000_000n,
  operatorStake: 1_000_000_000_000_000_000n,
  ageInDays: 30,
  feedbackCount: 0,
  averageFeedbackScore: 0,
}

const score = calculateTrustScore(input)
// { normalized: 72, tier: 'high', components: {...}, confidence: 0.6 }

Capabilities

import { CapabilitySchema, CapabilityCategoryEnum } from '@agentids/schema'

const cap = CapabilitySchema.parse({
  name: 'code-generation',
  description: 'Generates code review comments',
  category: 'code-generation',
})

Predicates

import { PREDICATES } from '@agentids/schema'

// Available predicates for Intuition Triples:
// 'has-capability' | 'operated-by' | 'endorses' | 'reports-issue'

Exports

| Export | Description | |---|---| | AgentRegistrationSchema | Full ERC-8004 registration with AgentID extensions | | EndpointSchema | Agent communication endpoint (MCP, A2A, web, email) | | CapabilitySchema | Agent capability with category and optional pricing | | CapabilityCategoryEnum | 10 standard capability categories | | TrustScoreSchema | Computed trust score shape | | calculateTrustScore() | Calculate trust from on-chain vault data | | TRUST_WEIGHTS | Weight config for score components | | TIER_THRESHOLDS | Score thresholds for tier classification | | PREDICATES | Standard predicate labels for Intuition Triples |

Trust Tiers

| Tier | Score | Recommendation | |---|---|---| | Elite | 85-100 | Highly trusted, suitable for critical tasks | | High | 65-84 | Trusted for most tasks | | Medium | 40-64 | Use for low-stakes tasks, verify outputs | | Low | 0-39 | Exercise caution, new or disputed agent |

License

MIT