Human verification for AI agents — proof of humanity, sybil resistance, and 'is
Human verification for AI agents — proof of humanity, sybil resistance, and 'is this wallet a real person' answers in one paid call. Returns two ZK-style outputs about any EVM wallet: (1) total humanity score 0-255 (15 = baseline reCAPTCHA, 45 = + Google OAuth, 70 = + Apple ID, 105 = + 2FA, 255 = full stack), with optional `min_score=N` for a yes/no `passes` verdict; (2) per-method verification check via `check=telegram,apple,g2fa,...` — buyer must NAME the methods to ask about, we never enumerate the full set. Backed by 141,871 Twin Matrix Soulbound Tokens on BNB Chain plus the canonical humanity registry off-chain. The underlying verification list never leaves Twin3 — only verdicts.
Paid test badge: not yet. The checks above are free: we call the tool without paying and read the payment request it sends back. The Verified badge needs paid calls whose answers match the promised output, and nobody can buy a badge.
Endpoint
GET https://human.twin3.ai/v1/human
| Category | Company and compliance |
|---|---|
| Provider host | human.twin3.ai |
| Networks | eip155:196, eip155:8453 |
| Payment schemes | exact |
| Self-reported calls, 30 days | 218 from 6 payers (the provider's figure, not ours) |
Our checks, last 30 days
| Day | Result | HTTP | Asked | Time |
|---|---|---|---|---|
| 2026-09-24 | valid payment request | 402 | $0.001 | 415 ms |
Example input (from the provider)
{
"method": "GET",
"queryParams": {
"check": "telegram,apple",
"include": "verified_at,dimensions_filled",
"match": "age:25-34,interest:travel,income:200-500+",
"min_score": "70",
"wallet": "0x1234567890123456789012345678901234567890"
},
"type": "http"
}
Promised output schema (from the provider)
{
"$schema": "https://json-schema.org/draft/2020-12/schema",
"properties": {
"input": {
"additionalProperties": false,
"properties": {
"method": {
"enum": [
"GET"
],
"type": "string"
},
"queryParams": {
"properties": {
"check": {
"description": "Optional CSV of specific verification methods to check (minimal-disclosure: buyer must name them; we don't enumerate). Returns `verifications: {method: bool}` for ONLY the methods asked. Valid methods: recaptcha, google, discord, line, telegram, apple, g2fa, onchain, worldid.",
"type": "string"
},
"include": {
"description": "Optional CSV of richer fields: `verified_at` (SBT mint timestamp, useful for sybil-pattern detection), `dimensions_filled` (count of non-zero matrix bytes, profile completeness signal).",
"type": "string"
},
"match": {
"description": "Optional CSV of attribute match conditions `key:value` (AND-ed). Returns pass/fail per condition + overall `matched` \u2014 NEVER the raw values (minimal disclosure). Keys: age, gender, occupation, income, social, web3, ai, risk, interest. A `+` suffix on ordered keys (age/income/web3/ai/risk) means '\u2265 this level'. Examples: `age:25-34`, `income:200-500+`, `social:instagram`, `interest:travel`.",
"type": "string"
},
"min_score": {
"description": "Optional integer 0-255. When set, response adds `passes: bool` indicating whether the wallet's humanity score meets the threshold. Reference levels: 15 = baseline (reCAPTCHA + sybil filter), 45 = + Google OAuth, 70 = + Apple ID, 105 = + 2FA, 255 = full verification stack.",
"pattern": "^([0-9]|[1-9][0-9]|1[0-9][0-9]|2[0-4][0-9]|25[0-5])$",
"type": "string"
},
"wallet": {
"description": "EVM address (0x-prefixed, 40 hex chars). The wallet to verify.",
"pattern": "^0x[a-fA-F0-9]{40}$",
"type": "string"
}
},
"required": [
"wallet"
],
"type": "object"
},
"type": {
"const": "http",
"type": "string"
}
},
"required": [
"type",
"method"
],
"type": "object"
},
"output": {
"properties": {
"example": {
"type": "object"
},
"type": {
"type": "string"
}
},
"required": [
"type"
],
"type": "object"
}
},
"required": [
"input"
],
"type": "object"
}