Faraday

​

Overview

​

The Problem with Traditional Static Analysis

• Rigid AST parsing with predefined pattern matching
• Hardcoded rules that generate high false positive rates
• No business logic understanding or contextual awareness
• Pattern-only detection that misses novel vulnerability patterns

​

The Faraday Breakthrough: Agent-Based Intelligence

​

Multi-Agent Architecture

​

1. Vulnerability Scanner Agent

• Intra-file analysis: Single-file vulnerability detection
• Inter-file analysis: Cross-file data flow and import tracking
• Architectural analysis: System-wide security patterns
• Hybrid approach: Traditional taint analysis + AI semantic reasoning

// Traditional scanners miss this - no direct user input
const config = loadConfig(); // reads from env/file
const query = `SELECT * FROM ${config.tableName}`; // SQL injection if config is tainted

// Faraday traces back to config source and flags the vulnerability

​

2. Context Validator Agent

• Uses file system tools (grep, readFile, find) to cross-reference findings
• Verifies vulnerabilities against actual code evidence
• Understands sanitization, validation, and protection layers
• Eliminates 90%+ of false positives

Finding: Possible SQL injection in getUserData()
Context Validator checks:
  ✓ Input validation present in middleware
  ✓ Parameterized query used
  ✓ ORM with built-in escaping
Result: False positive - marking as safe

​

3. Scoring Agent

• Assigns CVSS scores based on real exploitability
• Calculates attack vectors and complexity
• Considers authentication requirements
• Evaluates business impact context

• Is the vulnerable function user-reachable?
• What authentication is required?
• What data is at risk?
• How complex is exploitation?

​

4. Remediation Agent

• Generates production-ready patches with precise line boundaries
• Understands code style and project conventions
• Creates minimal, surgical fixes
• Provides detailed fix explanations

- const query = `SELECT * FROM users WHERE id = ${userId}`;
+ const query = 'SELECT * FROM users WHERE id = ?';
+ db.query(query, [userId], callback);

​

How Agent Collaboration Works

​

The AI Advantage

• Understand code intent beyond syntax
• Reason about business logic and security implications
• Detect novel patterns that rule-based scanners miss
• Provide context-aware recommendations

​

Intelligence Over Rules

// Traditional scanner: ❌ Flags as SQL injection (false positive)
// Faraday: ✅ Understands ORM abstraction provides protection

const user = await User.findOne({
  where: { email: req.body.email } // Sequelize ORM with parameterization
});

​

Multi-File Vulnerability Detection

// auth.js
export function getToken(req) {
  return req.headers.authorization; // Source
}

// middleware.js  
import { getToken } from './auth.js';
export function validateUser(req) {
  const token = getToken(req); // Flow
  return jwt.verify(token, secret); // Sink - but safe
}

// Faraday understands the complete flow and validates safety

​

Performance & Results

​

Scan Performance

​

Real-World Impact

Scan complete: 1,247 findings
├─ 892 false positives (71%)
├─ 312 low severity (25%)
└─ 43 actionable (4%)
Developer time: 8 hours triaging

Scan complete: 67 findings
├─ 3 false positives (4%)
├─ 52 with automated patches (78%)
└─ 67 actionable (100%)
Developer time: 45 minutes reviewing + applying fixes

​

Running Scans

​

Repository Scan

vulnzap scan --help

1. Vulnerability Scanner analyzes all code
2. Context Validator verifies findings
3. Scoring Agent prioritizes risks
4. Remediation Agent generates patches

​

Agent-Aware Development

​

MCP Protocol Integration

• Cursor
• Windsurf
• Cline
• VS Code (with MCP extension)

User: "Create a file upload endpoint"

AI generates code → Commits locally -> Initiates a commit scan -> Results given to agent -> Provides secure version

✅ Path traversal protection added
✅ File type validation included  
✅ Size limits enforced
✅ Malware scanning hook added

​

Why Faraday Changes Everything

​

Traditional Approach

Rules → AST → Pattern Match → Report → Manual Fix
⚠️ High false positives, no context, manual remediation

​

Faraday Approach

AI Agents → Multi-layer Analysis → Context Validation → Auto-remediation
✅ Low false positives, full context, automated fixes

​

Key Differentiators

​

The Bottom Line

• 5-7 minute full repository scans
• <5% false positive rate
• Automated remediation with working patches
• Multi-file vulnerability detection traditional scanners can’t match

​

Next Steps