7-Engine Architecture

Agnech uses a sophisticated multi-engine approach for comprehensive security analysis. Each engine specializes in different vulnerability detection techniques.

Overview

The 7-engine architecture provides defense in depth - each engine catches vulnerabilities that others might miss. Findings confirmed by multiple engines receive higher confidence scores.

Engines

96.1%

Accuracy

<4%

False Positives

2.8K+

Patterns

Engine Architecture

Engines are organized into two phases: Quick Scan (parallel, fast) and Deep Scan (sequential, thorough).

// Engine Execution Flow

Quick Scan:Pattern+SAST+CVE+Graph(parallel)

Deep Scan:Taint→Symbolic→AI(sequential)

Multi-Engine Consensus

When multiple engines detect the same vulnerability, the confidence score is boosted by +8%. This reduces false positives and highlights high-priority issues.

Quick Scan Engines

These engines run in parallel during every scan. They're optimized for speed while maintaining high accuracy.

Pattern Matching

Fast regex-based detection using 2,847+ vulnerability patterns

<1s

SAST Analysis

Static Application Security Testing with AST-based code analysis

2-5s

Dependencies & CVE

Scans dependencies against 23,847+ known CVEs

3-8s

Graph Analysis

Control flow and call chain tracking for inter-file vulnerabilities

2-5s

Deep Scan Engines

These engines run sequentially during deep scans. They provide thorough analysis for complex vulnerabilities.

Taint Analysis

Tracks untrusted data flow from sources to sinks

5-10s

Symbolic Execution

Mathematical proofs for integer overflow and edge cases

30-60s

AI Verification

Neural network verification to reduce false positives

5-10s

Engine Comparison

Compare all engines side by side. Click on any engine to see detailed capabilities.

7-Engine Architecture

Sort by:

Engine	Time	Accuracy	False Positives	Languages
Pattern	<1s	92%	8%	TypeScript, JavaScript +4
SAST	2-5s	95%	5%	TypeScript, JavaScript +1
CVE	3-8s	99%	1%	All
Taint	5-10s	94%	6%	TypeScript, JavaScript +1
Symbolic	30-60s	98%	2%	TypeScript, JavaScript +1
Graph	2-5s	93%	7%	TypeScript, JavaScript +1
AI	5-10s	96%	4%	All

Included in Deep Scan

Active Engine

See it in Action

Watch a deep scan with all 7 engines. Click play to start the demo.

bloodhound --deep ./src