Checking User-Thrown Gates

Why this skill is separate from executing-plans

User-gate enforcement is an opt-in flow. When the opt-in hook is not registered, executing-plans runs unchanged — no extra checks, no extra context, no extra questions. When the hook IS registered, it routes user-gate tasks through this skill. Keeping the decision logic in a separate skill means:

• Users who don't want the flow get zero friction.
• Users who do want it get a focused, scoped handler.
• executing-plans stays short and readable.

When to invoke

Any one of:

1. You are about to start a task whose json:metadata has "userGate": true or whose tags contains "user-gate" AND the opt-in hook is active (see the README for how to detect this — if you were invoked via /gate-check, the hook is active by definition).
2. A hook fired stderr telling you to run /gate-check <task-id>.
3. The user manually ran /gate-check <task-id>.

If none of these apply, return to executing-plans without running this skill.