Skill Instructions

Inputs / Outputs / Non-goals

• Inputs: feature requirements, settings defaults, UI + engine touchpoints.
• Outputs: flag definitions, guarded logic, observability notes, test updates.
• Non-goals: shipping unflagged behavior or changing public APIs silently.

Trigger conditions

Use this skill when prompts include or imply:

• Introducing new gameplay features.
• Running experiments or phased rollouts.
• Refactoring risky behavior that should remain reversible.

Mandatory rules

• New features default to OFF.
• Declare flags centrally in settings defaults.
• Guard both UI and engine behavior behind the same flag.
• Expose flag state via data-* attributes or equivalent observability hook.
• Preserve existing behavior when a flag is disabled.

Validation checklist

• Flag added with default OFF and documented usage.
• On/off behavior covered by targeted tests.
• Core checks run: npm run check:jsdoc && npx prettier . --check && npx eslint . && npm run check:contrast.
• Dynamic import and console discipline checks run.

Expected output format

• Flag definition details (name, default, scope).
• Guarded code paths (UI + engine) and observability notes.
• Validation evidence including targeted on/off test commands.

Failure/stop conditions

• Stop if rollout cannot be made reversible.
• Stop if requested behavior bypasses centralized flag declaration.