Skip to content

runbooks inventory validate-mcp

Auto-generated from runbooks inventory validate-mcp --help on 2026-05-21. Source of truth: runbooks PyPI package v1.3.22

Usage: runbooks inventory validate-mcp [OPTIONS]

  MCP cross-validation framework for data accuracy (≥99.5% target).

  🔄 4-WAYS VALIDATION WORKFLOW
  ├── Phase 1: MCP Discover
  │   └── Query awslabs.core-mcp for resource counts
  ├── Phase 2: Forward Validation
  │   └── Runbooks CLI → MCP (verify runbooks outputs)
  ├── Phase 3: Backward Validation
  │   └── MCP → Runbooks CLI (verify MCP consistency)
  └── Phase 4: Ground Truth Validation
      └── Direct AWS API calls for final verification

  ✅ ACCURACY TARGETS
  • Overall: ≥99.5% agreement across all phases
  • Per-phase status: Phase 1 LIVE, Phase 2 DEFERRED, Phase 3 DEFERRED, Phase 4 LIVE
  • Validation modes: MCP servers + Direct AWS APIs

  🎯 STATUS
  • Framework: src/runbooks/inventory/mcp_validator.py
  • Phase 1 (count): LIVE — uses Resource Explorer real data
  • Phase 2 (cost): DEFERRED — MCP cost server not connected
  • Phase 3 (consistency): DEFERRED — MCP inventory server not connected
  • Phase 4 (ground truth): LIVE — Direct AWS API validation

  Examples:
      # Validate EC2 inventory
      runbooks inventory validate-mcp --resource-type ec2 --profile CENTRALISED_OPS_PROFILE

      # Validate Lambda with custom threshold     runbooks inventory validate-
      mcp --resource-type lambda --threshold 95.0

      # Validate VPC with larger sample     runbooks inventory validate-mcp
      --resource-type vpc --sample-size 20 --output /tmp/vpc-validation.json

  📖 MCP validation ensures ≥99.5% accuracy across inventory discovery
  workflows

Options:
  --resource-type TEXT   AWS resource type to validate (e.g., ec2, lambda,
                         vpc)  [required]
  --profile TEXT         AWS profile for validation operations
  --output PATH          Path to save JSON validation results
  --sample-size INTEGER  Number of resources for ground truth sampling
  --threshold FLOAT      Minimum accuracy threshold (default: 99.5%)
  --help                 Show this message and exit.

Examples