🛡️ SCAM AI - AI-Powered Solana Security Scanner

Advanced security scanner powered by Machine Learning for detecting rug pulls and scams on Solana tokens (Pump.fun & DexScreener).

🤖 Powered by Artificial Intelligence

Our scanner uses a Machine Learning model trained on 559+ real Solana tokens to predict rug pulls with 83.04% accuracy.

📈 AI Model Statistics

Training Dataset: 559 tokens (318 RUG, 241 SAFE)
Model Type: Random Forest Classifier
Accuracy: 83.04%
Precision: 85.7%
Features Analyzed: 61+ on-chain metrics per token
Constantly Evolving: Model retrains automatically as new data is collected

🧠 What Makes Our AI Special?

✅ Real-World Training: Trained on actual rug pulls and successful tokens from Pump.fun ✅ 61+ Features: Analyzes holder distribution, price patterns, liquidity, creator history, and more ✅ Continuous Learning: Dataset grows daily through automated collection ✅ Dual-Model Approach: Random Forest + Gradient Boosting for maximum accuracy ✅ Confidence Scoring: Tells you how confident the AI is in its prediction

🚀 Features

🎯 Core Detection Systems

1. AI/ML Prediction

Machine Learning model trained on 500+ tokens
Predicts RUG/SAFE/SUCCESS with confidence score
AI Score: 0-100 (higher = safer)
Real-time prediction on every scan

2. Creator Analysis

Serial rugger detection (15+ tokens = EXTREME warning)
Rug percentage analysis (80%+ = confirmed scammer)
All previous tokens tracked
Automatic rug detection based on:
- Near-zero market cap
- Liquidity removal
- Bonding curve crashes
- Price dumps >90%

3. Whale Dumping Detection

Detects when top holders are selling
Coordinated exit warnings (3+ holders selling)
Exchange detection in top 10
Top 3 concentration analysis
Holder age tracking

4. Advanced Pump & Dump Analysis

Pattern type detection:
- Fast pump slow dump
- Coordinated pumps
- Dead cat bounce
- Stairs pattern (bot trading)
Pump/dump speed calculation
Manipulation confidence score (0-100%)

5. Analysis Confidence Scoring

Confidence score: 0-100%
Confidence levels: VERY LOW / LOW / MEDIUM / HIGH / VERY HIGH
Factors affecting confidence:
- Token age
- Data completeness
- Analyzer availability
Dynamic color-coded display (RED/ORANGE/GREEN)

6. Traditional Detection

Fake Holders Detection - Sybil attacks and bot wallets
Insider Trading Analysis - Snipers and coordinated buying
Wash Trading Scanner - Fake volume detection
Social Media Verification - Twitter, Telegram, Website checks
Real-Time Blockchain Data - Live Solana RPC analysis

🎨 Web Interface

Beautiful terminal-style interface with:

🔴 Red Theme: Cybersecurity aesthetic
🤖 AI Prediction Banner: ML predictions displayed prominently
📊 Risk Scores: 0-100 comprehensive scoring
🚨 Red Flags: Detailed warnings and explanations
💡 Recommendations: Actionable advice
📈 Confidence Indicators: Know how reliable the analysis is
⚡ Real-Time Scanning: Instant results

🤖 Discord & Telegram Bots

Scan tokens directly from Discord or Telegram!

Discord Bot: Slash commands with beautiful embeds
Telegram Bot: Simple commands with formatted messages
Real-Time Analysis: Powered by the same AI engine
Easy Setup: Just add your bot token and run

📖 See README_BOTS.md for setup instructions

Quick Start Bots

# Install bot dependencies
pip install discord.py python-telegram-bot

# Configure tokens in .env
DISCORD_BOT_TOKEN=your_token
TELEGRAM_BOT_TOKEN=your_token

# Run Discord bot
python discord_bot.py

# Run Telegram bot
python telegram_bot.py

🔧 Quick Start

Installation

# Clone the repository
git clone https://github.com/Cooktrenches/Scan-Scam-AI-v1.git
cd "Scan-Scam-AI-v1"

# Install dependencies
pip install -r requirements.txt

# Start web server
python web_app.py

Access at: http://localhost:5000

Using the Scanner

Go to http://localhost:5000
Paste a Solana token mint address
Click "Scan Token"
Get instant AI-powered analysis!

📊 Understanding the Scores

Risk Score (0-100)

Score	Level	Color	Description
0-24	LOW	🟢 Green	Relatively safe, low risk signals
25-49	MEDIUM	🟡 Yellow	Proceed with extreme caution
50-74	HIGH	🔴 Red	Very dangerous, multiple red flags
75-100	EXTREME	⛔ Dark Red	DO NOT BUY - Almost certain rug

AI Score (0-100)

Score	Prediction	Meaning
0-30	RUG	AI predicts this is a rug pull
30-70	SAFE	AI predicts moderate safety
70-100	SUCCESS	AI predicts high potential success

Confidence Score (0-100)

Score	Level	Color	Meaning
0-30	VERY LOW	🔴 Red	Limited data, take with caution
30-50	LOW	🔴 Red	Some missing data
50-70	MEDIUM	🟡 Orange	Good data coverage
70-85	HIGH	🟢 Green	Comprehensive analysis
85-100	VERY HIGH	🟢 Green	Full data, very reliable

🧪 ML Model Training (Advanced)

Want to improve the AI yourself? You can retrain the model with new data:

Automatic Data Collection

cd ml_module

# Collect 500 tokens from DexScreener (auto-labeled)
python collect_from_dexscreener_simple.py 500

# Retrain the model
python train_with_my_data.py

Manual Training

cd ml_module

# View current dataset
python show_features.py

# Train from existing CSV
python train_from_csv.py

The model automatically:

Extracts 61+ features per token
Labels tokens as RUG/SAFE based on market behavior
Trains Random Forest and Gradient Boosting models
Saves the best performing model
Updates the web scanner instantly

🔬 Technical Details

Architecture

Backend: Flask (Python)
ML Framework: Scikit-learn
Blockchain: Solana Web3.py + RPC
APIs: Pump.fun, DexScreener, Helius
Frontend: Vanilla JS + CSS (terminal theme)

ML Pipeline

Data Collection: Automated scraping from DexScreener
Feature Extraction: 61 metrics from blockchain + market data
Auto-Labeling: Smart heuristics to label RUG vs SAFE
Training: Dual models (Random Forest + Gradient Boosting)
Evaluation: Cross-validation with precision/recall metrics
Deployment: Automatic model updates without downtime

Analyzed Metrics (61+ Features)

Holder distribution patterns
Top holder concentration
Whale selling behavior
Fresh wallet detection
Sniper activity
Wash trading indicators
Price volatility patterns
Liquidity metrics
Volume analysis
Creator token history
Social media presence
Blockchain metadata
And much more...

📈 Roadmap

Current Version (v1.0 BETA)

✅ ML model trained on 559 tokens ✅ 83% accuracy ✅ Web interface ✅ Real-time scanning ✅ Advanced detection systems ✅ Confidence scoring

Future Updates

🔜 v1.1: Expand dataset to 1,000+ tokens 🔜 v1.2: Add historical price chart analysis 🔜 v1.3: Multi-chain support (Ethereum, BSC) 🔜 v1.4: API endpoints for developers 🔜 v1.5: Mobile app

⚠️ Important Disclaimers

Educational Purposes Only: This tool is designed for research and education.

Not Financial Advice: Never invest based solely on automated analysis.

No Guarantees: Even with 83% accuracy, the model can make mistakes.

DYOR: Always Do Your Own Research before investing.

Alpha Software: This is BETA software. Use at your own risk.

🤝 Contributing

We welcome contributions! The model improves as the dataset grows.

How to Help

Report False Positives/Negatives: Help us improve accuracy
Contribute Data: Run collection scripts and share labeled data
Improve Detection: Add new features or detection methods
Documentation: Help improve guides and tutorials

📞 Support

GitHub: Cooktrenches

📝 License

MIT License - See LICENSE for details.

🙏 Acknowledgments

Solana Foundation for blockchain infrastructure
Pump.fun for revolutionizing token launches
DexScreener for market data API
The entire Solana community

⚡ Built with AI, Trained by Real Data, Powered by the Community

Made with ❤️ for the Solana ecosystem

Name		Name	Last commit message	Last commit date
Latest commit History 4 Commits
ml_module		ml_module
static		static
templates		templates
.env.example		.env.example
.gitignore		.gitignore
DEPLOY_DISCORD_BOT_RENDER.md		DEPLOY_DISCORD_BOT_RENDER.md
README.md		README.md
README_BOTS.md		README_BOTS.md
README_WEB.md		README_WEB.md
START_DISCORD_BOT.bat		START_DISCORD_BOT.bat
START_TELEGRAM_BOT.bat		START_TELEGRAM_BOT.bat
START_WEB.bat		START_WEB.bat
analyser.bat		analyser.bat
authority_checker.py		authority_checker.py
batch_analyzer.py		batch_analyzer.py
config.py		config.py
creator_checker.py		creator_checker.py
database.py		database.py
debug_dexscreener.py		debug_dexscreener.py
deploy_pythonanywhere.sh		deploy_pythonanywhere.sh
dex_api.py		dex_api.py
discord_bot.py		discord_bot.py
helius_api.py		helius_api.py
insightx_api.py		insightx_api.py
installer.bat		installer.bat
liquidity_analyzer.py		liquidity_analyzer.py
main.py		main.py
metadata_fetcher.py		metadata_fetcher.py
onchain_analyzer.py		onchain_analyzer.py
pump_dump_detector.py		pump_dump_detector.py
render.yaml		render.yaml
requirements.txt		requirements.txt
risk_scorer.py		risk_scorer.py
sniper_detector.py		sniper_detector.py
social_checker.py		social_checker.py
solscan_api.py		solscan_api.py
start.sh		start.sh
stats.py		stats.py
telegram_bot.py		telegram_bot.py
test_blockchain_metadata.py		test_blockchain_metadata.py
test_detection_demo.py		test_detection_demo.py
test_full_scan.py		test_full_scan.py
test_insight_api.py		test_insight_api.py
test_insightx.py		test_insightx.py
test_insightx_correct.py		test_insightx_correct.py
test_pumpfun_metadata.py		test_pumpfun_metadata.py
test_solscan.py		test_solscan.py
test_token.bat		test_token.bat
test_token.py		test_token.py
test_volume_analyzer.py		test_volume_analyzer.py
test_volume_data.py		test_volume_data.py
token_analyzer.py		token_analyzer.py
top_holders_analyzer.py		top_holders_analyzer.py
vercel.json		vercel.json
volume_analyzer.py		volume_analyzer.py
wallet_analyzer.py		wallet_analyzer.py
web_app.py		web_app.py
wsgi.py		wsgi.py

Cooktrenches/Scan-Scam-AI-v1

Folders and files

Latest commit

History

Repository files navigation