@the-governor-hq/constitution-wearables

AI Safety Constitution for Wearable/Smartwatch Data Projects

AI behavior guidance layer working with consumer wearable health data (Garmin, Apple Watch, Whoop, Oura, Fitbit, etc.). Prescriptive, executable constraints that prevent medical claims and ensure ethical boundaries.

Quick Start

npm install --save-dev @the-governor-hq/constitution-wearables

Then run setup to configure your AI tools:

npx governor-install

This creates:

• .cursorrules — Safety rules for Cursor AI
• .vscode/settings.json — Copilot instructions
• .mcp-config.json — MCP server for Claude/ChatGPT
• package.json — Adds ai:context script

What It Does

When added to your AI agent's context, this framework:

• ✅ Prevents medical claims in generated code and text
• ✅ Enforces baseline requirements before recommendations
• ✅ Controls language patterns to avoid liability
• ✅ Defines clear scope boundaries for wearable data systems
• ✅ Protects all stakeholders — users, developers, and products

Wearable Data Types Covered

• Sleep tracking - Duration, stages, quality metrics
• Heart rate variability (HRV) - Recovery and stress signals
• Resting heart rate - Baseline cardiovascular data
• Activity tracking - Steps, distance, intensity
• Training load - Workout impact and recovery needs
• Readiness scores - Multi-signal wellness indicators
• Stress metrics - Physiological stress markers
• VO2 Max estimates - Cardio fitness proxies

Using MCP (Claude Desktop, etc.)

npm run ai:context

Core Principles for Wearable Data

| Principle | Detail | |---|---| | Personal baseline | Systems must learn each user's normal over time (30–90 days) | | Deviation-driven | Recommendations only when meaningful change is detected | | Behavioral suggestions | Timing adjustments, rest cues, activity modifications — not medical interventions | | Non-medical | No diagnoses, no supplements, no treatment protocols, no disease names | | Optionality | "Consider" and "might help" — never "you must" or "you should" | | Safety first | When in doubt about a feature, default to NO until confirmed safe |

Documentation

• 📖 Full Documentation
• 🐙 GitHub Repository
• 💬 Discussions

Related Packages

• @the-governor-hq/constitution-core - Core safety infrastructure (auto-installed)
• @the-governor-hq/constitution-bci - Brain-computer interface data
• @the-governor-hq/constitution-therapy - Therapy and mental health data

License

MIT