Ops Identity Hygiene

OpsIdentity Operational Hygiene Skill

Improve OpsIdentity code to detect administrative disorder, architectural debt, and configuration drift in Active Directory environments.

"Not looking for Russian hackers. Looking for administrative chaos, poor architecture, and suboptimal configurations."

When to Use This Skill

Mission

This skill focuses on operational hygiene, not penetration testing:

Stack Overview

• Frontend: React + TypeScript + Vite + TailwindCSS + shadcn/ui
• Backend: Node.js + Express + PostgreSQL
• Collection: PowerShell embedded in client/src/pages/NewAssessment.tsx
• AI: Anthropic Claude (Opus 4.5 for complex, Sonnet 4.5 for standard)
• Anti-Hallucination: Smart Filtering + Grounding Validation + Deterministic Rules

Key Files

Workflow

Phase 1: Identify Coverage Gap

Check current coverage (37% of 87 metrics). Reference 📊 Coverage Matrix for:

• Missing metrics by category
• Priority level (Critical/High/Medium)
• Implementation complexity

Phase 2: Implement PowerShell Function

Add collection function in NewAssessment.tsx. All functions must follow anti-null pattern:

function Get-NewMetric {
    $results = @()  # Always initialize as empty array
    
    try {
        $items = @(Get-ADObject -Filter * -ErrorAction Stop)  # Force array
        
        foreach ($item in $items) {
            try {
                $obj = @{
                    Name = $item.Name
                    IsProblematic = $false
                    # Add detection logic
                }
                $results += $obj
            } catch {
                Write-Host "[!] Error processing $($item.Name): $_" -ForegroundColor Yellow
            }
        }
        
        return @($results)  # Always return array
    } catch {
        Write-Host "[!] CRITICAL: $_" -ForegroundColor Red
        return @()  # Never return null
    }
}

Phase 3: Add Smart Filtering

Update filterCategoryData() in server.js to pre-filter problematic objects:

case 'NewCategory':
  return data.filter(item => 
    item.IsProblematic === true ||
    item.RiskLevel === 'HIGH' ||
    item.SpecificCondition > threshold
  );

Phase 4: Create AI Prompt

Add category instructions in server.js. See 📝 Prompt Templates.

Phase 5: Add Validation Rules

Update ATTRIBUTE_VALIDATION_RULES to prevent hallucinations:

'NEW_FINDING_TYPE': {
  category: 'NewCategory',
  identifierField: 'Name',
  validate: (obj) => obj.Condition === expectedValue
}

Phase 6: Increment Version

Update version in App.tsx after changes.

Critical Patterns

Anti-Null Guarantees

# ✅ CORRECT
$list = @()                              # Initialize empty
$items = @(Get-ADObject -Filter *)       # Force array context
return @($list)                          # Always return array

# ❌ WRONG
$list = $null                            # Never initialize as null
return $items                            # May return null if empty

Finding Output Format

{
  type_id: 'UPPERCASE_WITH_UNDERSCORES',
  title: '[Count] + specific problem',
  description: 'Current state + operational impact + compliance',
  recommendation: 'Copy-paste ready commands',
  affected_objects: ['REAL objects from JSON data']
}

What NOT to Do

• ❌ Focus on offensive security (Golden Ticket, lateral movement)
• ❌ Code without robust error handling
• ❌ Prompts that generate findings without real data
• ❌ Ignore anti-hallucination system
• ❌ Functions returning null instead of @()
• ❌ Unnecessary complexity

Valid Improvements

• ✅ Detect GPOs with 50+ settings (monolithic)
• ✅ Identify subnets not associated to any AD site
• ✅ Calculate replication latency between all DCs
• ✅ Find empty groups or groups without defined manager
• ✅ Detect users with estimated token size > 12KB
• ✅ Identify trusts with password > 90 days without rotation

Reference Files

Load these as needed during development:

• 🔍 Analyzer Prompt - Full analysis prompt with 87 metrics baseline, output format, web search rules
• 📊 Coverage Matrix - Current coverage status, gaps by category, sprint priorities
• 📝 Prompt Templates - AI prompt structure for new categories with examples
• ⚙️ PowerShell Functions - Ready-to-use function implementations

Quick Reference: 87 Industry Metrics

Current coverage: 37% → Target: 80%+