/run-everything

Toggles 'Run Everything' autopilot mode, which runs commands without asking for approval.

/run-everything (alias /auto-run) toggles Run Everything on/off, or shows the current state with status. In this mode the agent runs commands without individual approval prompts.

✅ WHEN TO USE IT?

Many repetitive, low-risk steps in a row

/run-everything on — e.g. for a long chain of lint fixes across many files

Check the current autopilot state before a task

/run-everything status — quick check before riskier steps come up

⛔ WHEN NOT TO?

Production or destructive commands are coming up

Without prompts, the agent can accidentally delete data or change production systems.

Better: Use /sandbox or /auto-review instead for controlled approvals

Only a single risky command needs to go through

Turning on the full autopilot for that opens up more than needed.

Better: Just approve the single prompt normally instead of granting blanket approval

SOURCES

Cursor Docs — Slash Commands ↗ cursor.com