Back to Projects
PythonTensorFlowSlitherMythril

AI Auditor

ML-based vulnerability detection system for Solidity contracts.

2024

Objective

Identify security flaws in Solidity contracts before deployment using ML.

Technical Challenges

  • 1.Limited labeled training data for smart contract vulnerabilities.
  • 2.High false-positive rates in static analysis tools.
  • 3.Capturing complex control flow patterns.

Outcomes & Impact

  • Achieved 87% accuracy in detecting critical vulnerabilities.
  • Reduced preliminary audit time by 60%.

Implementation

ai-auditor.sol
1# Vulnerability Detection Pipeline
2def analyze_contract(source_code):
3    ast = solidity_parser.parse(source_code)
4    cfg = control_flow_graph.generate(ast)
5    
6    # Extract features for Transformer model
7    features = feature_extractor.process(cfg, ast)
8    
9    # Predict vulnerability probability
10    prediction = model.predict(features)
11    
12    if prediction.score > THRESHOLD:
13        return Report(
14            type=prediction.class,
15            severity="CRITICAL",
16            location=prediction.line_no
17        )
18    return Report(status="SAFE")
Previous Project

DeFi Aggregator

View All Projects

Back to Portfolio