A CLI tool and AI skill that analyzes EVM token contracts for rug pull risks, honeypots, holder concentration, and smart money activity. Generates a 0-100 risk score with detailed flags.
Works with Binance Web3, OKX, OpenClaw, and any AI agent platform that supports CLI skills.
pip install requests
# Scan by contract address
python main.py --address 0x6982508145454ce325ddbe47a25d4ec3d2311933 --chain eth
# Scan by token name (auto-resolves address)
python main.py --token PEPE --chain eth
# JSON output for downstream processing
python main.py --address 0x... --chain bsc --jsonExample output:
========================================================
TOKEN SAFETY REPORT
Pepe (PEPE)
0x69825081...311933 [ETH]
========================================================
Score : 92/100 [##################--] SAFE
------------------------------------------------------
CONTRACT AUDIT
------------------------------------------------------
Open Source : [ OK ] Verified
Honeypot : [ OK ] No
Mintable : [ OK ] No
Blacklist : [!!] YES
Buy Tax : 0.0%
Sell Tax : 0.0%
------------------------------------------------------
HOLDER ANALYSIS
------------------------------------------------------
Total Holders : 508,775
Top 10 Hold : 41.5%
Insider Hold : 0.0%
LP Locked : [ OK ] Yes
------------------------------------------------------
FLAGS
------------------------------------------------------
[!] Blacklist function exists
========================================================
| Flag | Chain | Chain ID |
|---|---|---|
eth |
Ethereum | 1 |
bsc |
BNB Smart Chain | 56 |
polygon |
Polygon | 137 |
arbitrum |
Arbitrum One | 42161 |
base |
Base | 8453 |
avax |
Avalanche | 43114 |
op |
Optimism | 10 |
- Honeypot detection
- Source code verified / open source
- Mintable supply (owner can print tokens)
- Hidden owner / ownership backdoor
- Transfer pause function
- Blacklist / whitelist controls
- Buy and sell tax rates
- Selfdestruct function
- Cannot sell all tokens
- Total holder count
- Top 10 holder concentration %
- Insider / dev wallet holdings %
- LP locked status
- Whether the token appears in Binance's smart money inflow ranking
- 24h net inflow USD
- Buy / sell transaction counts from tracked wallets
| Score | Rating | Meaning |
|---|---|---|
| 80-100 | SAFE | Low risk |
| 60-79 | CAUTION | Some risks, proceed carefully |
| 40-59 | RISKY | Multiple flags, high caution |
| 0-39 | DANGER | Severe risks — honeypot or scam likely |
token-safety-scanner/
├── adapters/
│ ├── goplus.py # GoPlus Security API (free, no auth)
│ └── binance_web3.py # Binance Web3 public API (free, no auth)
├── scanner/
│ ├── contract_audit.py # Parses GoPlus data into structured audit
│ ├── holder_analysis.py # Concentration analysis from audit data
│ ├── whale_tracker.py # Smart money inflow via Binance
│ └── risk_scorer.py # Pure calculation, no API calls
├── main.py # CLI entry point
├── SKILL.md # AI agent skill definition
└── tests/
└── test_scanner.py
Data flow: main.py resolves address → calls contract_audit (GoPlus) → holder_analysis (from audit data) → whale_tracker (Binance) → risk_scorer (pure calc) → formats report.
| Source | Auth Required | Chains | Used For |
|---|---|---|---|
| GoPlus Security | None | 30+ | Contract audit, holder data |
| Binance Web3 API | None | ETH, BSC | Token search, smart money inflow |
No API keys needed to run this tool.
pip install pytest
python -m pytest tests/ -vTests use USDT on BSC as a known-safe token baseline.
This tool is designed to be called by AI agents. See SKILL.md for the full skill definition compatible with Binance Skills Hub.
The --json flag produces structured output suitable for agent consumption:
python main.py --address 0x... --chain bsc --json{
"address": "0x...",
"chain": "bsc",
"audit": { "is_honeypot": false, "buy_tax": 0.0, ... },
"holders": { "top10_concentration_pct": 26.8, ... },
"risk": { "score": 80, "rating": "SAFE", "flags": [...] }
}This tool is for informational purposes only. Data is sourced from third-party APIs and may be incomplete or inaccurate. This does not constitute financial or investment advice. Always do your own research (DYOR).
MIT