predictive_fraud¶

Forecasts the probability that a wallet address will engage in fraudulent activity before any transaction takes place. Also performs AML (Anti-Money Laundering) checks, returning detailed forensic signals alongside the score.

~98% accuracy on Ethereum
~96% accuracy on BNB Smart Chain
Not rules-based blocklisting - AI trained on 1.3B+ behavioural data points
Response latency under 100ms - suitable for real-time wallet-connect gating

MCP Endpoint: https://prediction.mcp.chainaware.ai/sse

Supported Networks¶

Identifier	Network
`ETH`	Ethereum
`BNB`	BNB Smart Chain
`POLYGON`	Polygon
`TON`	TON
`BASE`	Base
`TRON`	Tron
`HAQQ`	HAQQ

Input Schema¶

Field	Type	Required	Description
`apiKey`	string	Yes	Your ChainAware API key
`network`	string	Yes	One of the network identifiers above
`walletAddress`	string	Yes	Wallet address to evaluate (hex or ENS)

Output Schema¶

{
  "message": "string",
  "walletAddress": "string",
  "status": "Fraud | Not Fraud | New Address",
  "probabilityFraud": "0.00-1.00",
  "token": "string",
  "lastChecked": "ISO-8601 timestamp",
  "forensic_details": {
    "...": "deep on-chain behavioural metrics"
  },
  "createdAt": "ISO-8601 timestamp",
  "updatedAt": "ISO-8601 timestamp"
}

`status` Values¶

Value	Meaning
`Not Fraud`	Low probability - wallet appears safe
`Fraud`	High probability of fraudulent activity
`New Address`	Insufficient on-chain history to score - treat with caution

`probabilityFraud` Interpretation¶

Range	Risk Level	Recommended Action
`0.00-0.20`	Low	Safe to proceed
`0.21-0.50`	Medium	Proceed with caution; consider additional checks
`0.51-0.80`	High	Flag for manual review
`0.81-1.00`	Critical	Block or reject immediately

Code Examples¶

Node.js¶

const result = await client.callTool({
  name: "predictive_fraud",
  arguments: {
    apiKey: process.env.CHAINAWARE_API_KEY,
    network: "ETH",
    walletAddress: "0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045"
  }
});

// result.status           → "Not Fraud"
// result.probabilityFraud → "0.02"

Python¶

result = await session.call_tool("predictive_fraud", {
    "apiKey": os.environ["CHAINAWARE_API_KEY"],
    "network": "ETH",
    "walletAddress": "0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045"
})

# result["status"]           → "Not Fraud"
# result["probabilityFraud"] → "0.02"

Example Agent Prompts¶

"Is it safe to interact with vitalik.eth on Ethereum?"
"What is the fraud risk of 0xABC123... on BNB?"
"Run an AML check on this wallet: 0x456... on BASE"
"Screen all wallets connecting to my dapp today."
"Is this address on TRON flagged for suspicious behaviour?"

Use Cases¶

DeFi onboarding gate - screen wallets before allowing deposits, borrowing, or staking
Airdrop and incentive protection - block fraud-flagged wallets from claiming rewards before distribution
NFT launchpads - prevent fraudulent minters from participating in drops
AML compliance - satisfy KYT requirements at account creation or transaction initiation
Lending protocols - reject high-risk borrowers before loan origination
Continuous monitoring - re-check wallets periodically to catch behavioural changes over time

Error Codes¶

Code	Meaning
`403`	Invalid or missing `apiKey`
`400`	Malformed `network` or `walletAddress`
`500`	Temporary backend failure - retry after a short delay

predictive_behaviour - full behavioural profile, intent prediction, and personalisation; call this after predictive_fraud for a complete picture
predictive_rug_pull - use this for contract addresses instead of wallet addresses