How to identify phishing addresses?

Phishing addresses usually mimic well-known protocol addresses or contain special suffixes like dead, beef, cafe, etc. CryptoAgent Guard detects these characteristic patterns and compares them with known protocol whitelists, reminding you to verify address authenticity.

CryptoAgent Guard - AI Agent Transaction Instruction Local Security Audit

Name: CryptoAgent Guard
Author: CryptoAgent Guard

AI-generated JSON transaction objects or raw instructions

Security Score

Waiting for Audit

Risk Checklist

Contract Address Verification
Transaction Amount Check
Authorization Limit Review
Reentrancy Attack Detection
Contract Interaction Whitelist

Gas Estimate

Estimated Gas --

Gas Price --

Total Cost (ETH) --

Security Audit Rules

CryptoAgent Guard is a professional Local crypto transaction validator, providing reliable solutions for AI agent wallet safety check.

Contract Address Verification

As an Ethereum transaction payload inspector, verifies whether the target address conforms to the address format specification of the selected blockchain, detects zero addresses, black hole addresses and phishing characteristic addresses. Supports Ethereum EIP-55 checksum verification.

Transaction Amount Check

Monitors large asset transfers, intelligently identifies amount thresholds. Triggers warnings for amounts exceeding 10 ETH, marks as extremely high risk for amounts exceeding 100 ETH, preventing accidental large asset losses.

Authorization Limit Review

Detects high-risk functions such as approve, transferFrom, setApprovalForAll, verifies whether the authorization target is a known trusted protocol. Identifies unlimited authorization risks, ensuring AI agent wallet safety.

Slippage Verification (Reentrancy Attack Detection)

Checks whether slippage settings exceed the safety threshold (5%), preventing MEV attacks and sandwich attacks. Protects transactions from being sandwiched, ensuring AI agents execute expected operations.

Contract Interaction Whitelist

Compares target addresses with known DeFi protocol whitelists (including Uniswap, Aave, Compound, MakerDAO, etc.), identifying unknown contract interaction risks.

Frequently Asked Questions (FAQ)

How to locally verify crypto transaction security?

CryptoAgent Guard, as a professional Local crypto transaction validator, can complete transaction verification locally in the browser without RPC connections or third-party APIs. You just need to paste the transaction instructions generated by AI Agent to get a complete audit report including address verification, amount check, Gas estimation, etc., ensuring private keys never leave the local environment.

What is the Approve authorization risk?

The approve() function of ERC-20 allows a specified contract to withdraw a fixed quota of tokens on your behalf. If authorized to a malicious contract, attackers can use the unlimited authorization mechanism to transfer all assets. The AI agent wallet safety check module verifies whether the authorization target is a trusted protocol contract (such as Uniswap V2/V3 Router, Compound cToken, etc.) and issues high-risk warnings for unknown addresses.

Why is setApprovalForAll more dangerous?

Unlike approve, setApprovalForAll is used for ERC-721/1155 NFT authorization. Once authorized, the contract can transfer all NFTs you hold. This tool will mark this operation as "extreme risk" and attach a red warning border. It is recommended to use this authorization only for fully audited well-known NFT marketplaces.

Which blockchain networks does the tool support?

As a multi-chain Ethereum transaction payload inspector, CryptoAgent Guard supports: Ethereum Mainnet (EIP-155), Binance Smart Chain (BSC), Solana (Base58 address), Polygon, Arbitrum One. All chains include address format verification and Gas estimation functions.

How to perform Address checksum verification?

Ethereum addresses support EIP-55 mixed case checksums. This tool uses regex /^0x[a-f0-9]{40}$/i for basic format verification, detecting zero addresses (0x000...000) and black hole addresses (0x000...dead). Phishing addresses usually contain suspicious suffixes (dead, beef, cafe), which will be identified and warned.

Why do you need local audit instead of online tools?

Local audit achieves zero data transmission: your transaction instructions, wallet addresses, and signature information will never be sent to any server. This is particularly critical for protecting high-value wallets operated by AI Agents. All calculations (regex matching, Gas estimation, scoring) are completed in the WebAssembly/JavaScript sandbox.

How does Gas estimation work?

The tool automatically calculates based on transaction type recognition: ETH Transfer about 21,000 gas, ERC-20 Approve about 50,000 gas, Uniswap V2 Swap about 150,000 gas. Combines current network average Gwei (Ethereum 30 Gwei, BSC 5 Gwei) to calculate total costs, helping you evaluate transaction costs.

How is the audit score calculated?

Base score 100 points, deducting item by item based on detection results: black hole address -40, phishing address characteristics -30, authorization unknown contract -30, large transfer (>10 ETH) -15, unlimited authorization risk -20, slippage over 5% -10. Score ≥80 is safe, 60-79 is medium risk, 40-59 is high risk, <40 is extremely high risk.

Security Audit Methodology

1. Address Verification Mechanism

Ethereum Address Format Verification: Uses regex /^0x[a-f0-9]{40}$/i to validate hexadecimal address format.

Solana Address Format Verification: Uses regex /^[1-9A-HJ-NP-Za-km-z]{32,44}$/ to validate Base58 encoded addresses.

Black Hole Address Detection: Compares against known black hole address list (e.g., 0x000...0000, 0x000...dead).

Phishing Address Pattern Recognition: Detects suspicious suffixes at address end (dead, beef, cafe).

2. High-Risk Function Detection

Authorization Function Detection: Uses regex to match approve\(, transferFrom\(, setApprovalForAll\(, increaseAllowance\(, decreaseAllowance\(, safeTransferFrom\(.

Protocol Whitelist Verification: Compares authorization target address against known trusted protocol list (e.g., Uniswap, Aave, Compound).

Risk Assessment: If authorization target is not a known protocol, marks as extremely high risk and deducts corresponding points.

3. Transaction Amount Monitoring

Large Amount Warning Threshold: Triggers warning for amounts exceeding 10 ETH, marks as extremely high risk for amounts exceeding 100 ETH.

Value Parsing: Supports parsing value field in JSON and amounts in natural language descriptions.

Unit Recognition: Automatically recognizes ETH, BNB, SOL, wei, gwei and other units.

4. Gas Cost Estimation

Operation Type Recognition: Identifies operation types like Transfer, Swap, Approve, Contract Deployment based on input content.

Gas Baseline Values: Transfer costs ~21,000 gas, Swap costs ~120,000-150,000 gas, Approve costs ~45,000-50,000 gas.

Fee Calculation: Calculates estimated fee based on current network average Gwei price.

5. Scoring Algorithm

Base Score: 100 points

Deduction Items:

High-risk function detected but target is known protocol: -20 points
High-risk function authorized to unknown address: -50 points (-30 additional)
Invalid address format: -15 points
Black hole address detected: -40 points
Phishing address characteristics: -30 points
Large transaction (10-100 ETH): -15 points
Large transaction (>100 ETH): -35 points (-20 additional)
Slippage over 5%: -10 points
Unknown contract address: -10 points

Grade Classification:

SAFE: 80-100 points
WARNING (Medium Risk): 60-79 points
HIGH RISK: 40-59 points
CRITICAL (Extremely High Risk): 0-39 points