# Windows AI Controls

> Audit, disable, and restore selected Windows AI features through reviewable registry and policy changes with explicit elevation and verification.

- Skill: `batteryshark/windows-ai-controls` (Agent Skill, multi-file: 5 files)
- Install (CLI): `npx skillmds@latest add batteryshark/windows-ai-controls`
- Raw SKILL.md: https://api.skillmd.com/api/skills/batteryshark/windows-ai-controls/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Security
- Author: batteryshark (https://skillmd.com/u/batteryshark)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/batteryshark/windows-ai-controls

---


# Control Windows AI Features

1. Identify the Windows build, edition, managed-device status, and exact feature: Recall/snapshots, Copilot, Edge assistance, typing personalization, or an app-specific capability.
2. Run `bin/windows-ai-controls audit` without elevation and read [references/policy-safety.md](references/policy-safety.md).
3. Prefer documented policy controls for that build. Generate a plan showing every registry/policy value, previous value, scope, restart requirement, and restore action.
4. Obtain explicit approval before applying machine policy or removing a package. Use the narrowest elevation method and preserve a rollback record.
5. Re-run the audit after policy refresh/restart. Report unsupported or superseded settings instead of claiming success from a registry write alone.

This skill does not copy opaque third-party "debloat" scripts or bypass organization policy. Use [agents/policy-reviewer.md](agents/policy-reviewer.md) for a second review of a proposed change set.