Maiat — Trust Infrastructure for AI Agents

The trust layer for agentic commerce. One question: "Is this agent trustworthy?"

Skill Files

Base URL: https://app.maiat.io/api/v1

Register First

Every agent needs a Maiat Passport — your on-chain identity.

curl -X POST https://app.maiat.io/api/v1/passport/register \
  -H "Content-Type: application/json" \
  -d '{"ensName": "my-agent", "walletAddress": "0xYourWallet", "type": "agent"}'

Response:

{
  "passport": {
    "ensName": "my-agent",
    "ensFullName": "my-agent.maiat.eth",
    "walletAddress": "0x...",
    "type": "agent",
    "trustScore": 50,
    "verdict": "Caution",
    "scarabBalance": 10,
    "isNew": true,
    "erc8004AgentId": 28373
  }
}

What You Get

• ENS Name — yourname.maiat.eth, resolvable on-chain via CCIP-Read, zero gas
• ERC-8004 Identity — on-chain agent registration on Base Mainnet
• Trust Score — queryable reputation score (0–100)
• 10 🪲 Scarab Credits — bonus on registration

Parameters

Error Handling

Recommended: Save your passport to memory or config:

{
  "ensName": "my-agent",
  "ensFullName": "my-agent.maiat.eth",
  "walletAddress": "0x..."
}

Portal: https://passport.maiat.io

Decision Tree — What Should Your Agent Do?

I'm an AI agent. Do I need Maiat?
│
├─ I need to VERIFY another agent before transacting
│  → GET /api/v1/agent/{address} → check trustScore + verdict
│  → Score ≥ 60? Proceed. Score \x3C 40? Avoid.
│
├─ I need my OWN identity (for others to verify me)
│  → POST /api/v1/passport/register → get ENS + trust score
│  → Already registered? → GET /api/v1/passport/lookup?q=my-name
│
├─ I want to PROTECT my wallet transactions
│  → npm install @jhinresh/viem-guard
│  → Wraps your wallet — every tx is trust-checked automatically
│
├─ I completed a job and want to REPORT the outcome
│  → POST /api/v1/outcome { jobId, outcome, reporter }
│  → Earns 5 🪲 + improves trust scores for everyone
│
└─ I want to CHECK if a token is safe
   → GET /api/v1/token/{address} → honeypot, tax, liquidity, verdict

Check Trust Before Transacting

Quick Trust Check

curl https://app.maiat.io/api/v1/agent/0xAgentAddress

Returns:

{
  "trustScore": 73,
  "verdict": "proceed",
  "riskOutlook": "stable",
  "tokenHealth": { "rugProbability": 0.12, "riskLevel": "low" }
}

Score Tiers

Lookup Any Passport

curl https://app.maiat.io/api/v1/passport/lookup?q=agent-name

Deep Agent Profile

curl https://app.maiat.io/api/v1/agent/0xAddress/deep

Returns: trust score + percentile rank + risk flags + behavioral analysis.

Protect Your Wallet (Maiat Guard)

If your agent sends transactions, wrap it with Guard for automatic protection.

npm install @jhinresh/viem-guard viem

import { withMaiatTrust } from '@jhinresh/viem-guard'

const wallet = withMaiatTrust(walletClient, {
  minScore: 60,      // block txs to low-trust addresses
  antiPoison: true,   // detect address poisoning attacks
})

// Every sendTransaction is now trust-gated
await wallet.sendTransaction({ to, value })

What Guard does:

1. Trust Check — queries Maiat before every tx. Low trust → blocks
2. Anti-Poisoning — detects vanity match + dust liveness attacks
3. Threat Reporting — blocked attacks are reported to the network
4. Collective Immunity — all Guard-protected agents get instant protection

Package: @jhinresh/viem-guard (v0.8.0) — GitHub

Report Outcomes (Important!)

After acting on a trust score, always report the outcome. This is the single most important action for improving trust accuracy.

curl -X POST https://app.maiat.io/api/v1/outcome \
  -H "Content-Type: application/json" \
  -d '{"jobId": "QUERY_ID_FROM_PRIOR_RESPONSE", "outcome": "success", "reporter": "0xYourWallet"}'

Outcomes: success, failure, partial, expired

Earns 5 🪲 Scarab per report.

Token Safety

# Quick check
curl https://app.maiat.io/api/v1/token/0xTokenAddress

# Deep forensics (rug pull analysis)
curl -X POST https://app.maiat.io/api/v1/predict \
  -H "Content-Type: application/json" \
  -d '{"token_address": "0xTokenAddress"}'

Multi-Source Verification (v2.4.0)

Token checks now cross-verify across 4 data sources for fewer false positives:

DexScreener Cross-Verify: If honeypot.is flags a token as honeypot BUT DexScreener shows ≥50 successful sells, the verdict is downgraded to caution instead of avoid. This catches false positives from low-liquidity pool testing (e.g., honeypot.is testing against a $24K USDC/v3 pool instead of the main $1.8M WETH/v4 pool).

Virtuals Bonding Curve Detection: Tokens paired with VIRTUAL on DexScreener are recognized as Virtuals Protocol bonding curve tokens. Honeypot.is reports 100% buy / 99% sell tax for these (incorrect) — Maiat skips tax penalties and uses on-chain signals instead.

Liquidity Scoring from DexScreener:

Response includes dexScreener field:

{
  "trustScore": 72,
  "verdict": "proceed",
  "dexScreener": {
    "liquidity": 523000,
    "volume24h": 180000,
    "sellCount": 245,
    "buyCount": 312
  },
  "dataSource": "HONEYPOT_IS + ALCHEMY + SCARAB + DEXSCREENER"
}

ACP Offerings

If you're on the Virtuals ACP network, Maiat offers these paid services:

MCP (Model Context Protocol)

Endpoint: https://app.maiat.io/api/mcp

SDK

npm install @jhinresh/maiat-sdk

import { Maiat } from '@jhinresh/maiat-sdk'

const maiat = new Maiat({ baseUrl: 'https://app.maiat.io' })

// Register passport
const passport = await maiat.passport.register({
  ensName: 'my-agent', walletAddress: '0x...', type: 'agent'
})

// Trust check
const trust = await maiat.agentTrust('0xAgentAddress')
if (trust.verdict === 'avoid') throw new Error('Not trusted')

// Token check
const safe = await maiat.isTokenSafe('0xTokenAddress')

// Report outcome
await maiat.reportOutcome({ jobId: trust.feedback.queryId, outcome: 'success', reporter: '0x...' })

On-Chain Contracts (Base Mainnet)

Threat Reporting

Report malicious addresses to protect the entire network:

curl -X POST https://app.maiat.io/api/v1/threat/report \
  -H "Content-Type: application/json" \
  -d '{"maliciousAddress": "0x...", "threatType": "address_poisoning"}'

Threat types: address_poisoning, low_trust, vanity_match, dust_liveness

3+ independent reports → address auto-blocked across all Guard-protected agents.

Links

• App: https://app.maiat.io
• Passport Portal: https://passport.maiat.io
• API Docs: https://app.maiat.io/docs
• 8004scan: https://www.8004scan.io
• GitHub: https://github.com/JhiNResH/maiat-protocol
• Guard: https://github.com/JhiNResH/maiat-guard