Noise Reduction

Premium

Security scanners generate hundreds of findings. Noise reduction helps you focus on what matters by deduplicating findings and filtering out low-priority issues.

The Problem

100+

Findings per scan

80%

Are duplicates

Alert

Fatigue

How It Works

1. Fingerprinting

Every finding gets a unique, stable fingerprint based on its key attributes:

Scan Type	Fingerprint Components
SAST	rule_id + file_path + line_number
CSPM	check_id + resource_id + region
DAST	rule_id + URL + parameter
Runtime	title + resource_id + resource_type

2. Tracking States

Each finding can be in one of four states:

open→acknowledged→ignored→fixed

Acknowledged = tracked but won't alert | Ignored = false positive

Deduplication Commands

# View deduplication stats
$ codephreak dedup

   Deduplication Stats    
┏━━━━━━━━━━━━━━━┳━━━━━━━┓
┃ Status        ┃ Count ┃
┡━━━━━━━━━━━━━━━╇━━━━━━━┩
│ Open          │ 15    │
│ Acknowledged  │ 42    │
│ Ignored       │ 8     │
│ Fixed         │ 23    │
│ Total         │ 88    │
└───────────────┴───────┘

# Acknowledge a finding (won't alert, but still tracked)
$ codephreak dedup a1b2c3d4 --action acknowledge
✅ Acknowledged: a1b2c3d4

# Mark as false positive
$ codephreak dedup a1b2c3d4 --action ignore
🔇 Ignored: a1b2c3d4

# Reopen a finding
$ codephreak dedup a1b2c3d4 --action reopen
🔄 Reopened: a1b2c3d4

Noise Filter Configuration

Configure filtering rules in your codephreak.yml:

# codephreak.yml
noise_reduction:
  # Minimum severity to report
  min_severity: high    # Only critical + high
  
  # Paths to exclude
  exclude_paths:
    - node_modules/*
    - vendor/*
    - "*.test.js"
    - "*.spec.ts"
    - __tests__/*
  
  # Rules to exclude (too noisy or false positives)
  exclude_rules:
    - generic-api-key
    - hardcoded-password-default
    - insecure-random

Data Flow

Scan Results: 100 findings

↓ NoiseFilter

After filtering: 50 findings

removed low severity, test files

↓ DeduplicationEngine

New findings: 12

skipped acknowledged/ignored

Local Storage

Finding history is stored locally at:

~/.codephreak/findings_db.json

This allows deduplication to work across scans without requiring a server connection.