Digital Twin Auto

Automatically generate comprehensive digital twins of systems, processes, or business operations with calibrated parameters and continuous validation feedback loops.

When to Use This Command

Run this command when...

• You need to build a simulation model of a manufacturing process, customer journey, or system architecture for what-if analysis
• You want to create a virtual replica of an operational system that can be tested under stress conditions without production risk
• Your team needs a calibrated model that mirrors real-world behavior for training, forecasting, or optimization experiments

Do NOT use this command when...

• You need a real-time monitoring dashboard rather than a simulation model
• The system is too simple to justify a digital twin -- a spreadsheet model would suffice

Quick Start

Copy
# .claude/commands/digital-twin-auto.md
# Create digital twin automatically
Create digital twin for: $ARGUMENTS

Copy
# Run the command
claude "digital-twin-auto e-commerce checkout flow with payment gateway latency modeling"

Expected output:
- System component map with interfaces
- Parameterized simulation model
- Calibration results against historical data
- Validation metrics and accuracy scores
- What-if scenario testing interface

Core Concepts

Concept	Description
Twin Subject	The real-world system being replicated (process, asset, or workflow)
Parameter Calibration	Tuning model parameters to match observed real-world behavior
Validation Loop	Continuous comparison of twin outputs against actual outcomes
Interface Mapping	Defining how system components connect and exchange data
Fidelity Levels	Granularity tiers from abstract overview to high-fidelity replication

Digital Twin Architecture:

  Real System
       |
  [Data Collection]
       |
  [Parameter Extraction]
       |
  Twin Model
  |        |
  |   [Calibrate]<---+
  |        |         |
  |   [Validate]-----+
  |        |
  [Simulate Scenarios]
       |
  Insights & Predictions

Configuration

Parameter	Default	Description
Fidelity Level	Medium	Abstract, medium, or high-fidelity replication detail
Data Sources	Docs + logs	System documentation and historical data for calibration
Validation Threshold	80% accuracy	Minimum acceptable correlation with real-world outcomes
Update Frequency	On-demand	How often the twin re-calibrates against new data
Scenario Count	3	Number of what-if scenarios generated automatically

Best Practices

1. Start at medium fidelity -- build the twin at a manageable detail level first, then increase fidelity only where sensitivity analysis shows it matters
2. Supply historical data -- calibration quality depends on real observations. Include throughput numbers, latency measurements, or conversion rates in your arguments
3. Define system boundaries clearly -- specify exactly which components are inside and outside the twin to avoid modeling unnecessary complexity
4. Validate incrementally -- check each subsystem independently before validating the integrated twin to isolate calibration errors
5. Document assumptions -- every digital twin embeds simplifications. Record what you left out so future users know the model's limitations

Common Issues

1. Twin diverges from reality -- recalibrate with more recent data. Systems evolve, and parameter drift is the most common cause of twin inaccuracy
2. Model is too slow -- reduce fidelity in subsystems that contribute least to the metrics you care about. Not every component needs high-resolution modeling
3. Insufficient data for calibration -- use expert estimates as priors and flag the twin's confidence intervals as wider in those areas until real data becomes available