Operate
Approvals
Exposure determines what the model can propose. Approval determines which exposed effects may execute.
Modes
always-ask | Prompt before every effect classified as approval-required. |
write | Permit configured write behavior while retaining stronger command boundaries. |
yolo | Automatically approve exposed capabilities; safe only inside an external sandbox you trust. |
Inspect and change
Interactive
/approval status
/approval mode
/approval set write
/approval resetThe persistent default is tools.approvalMode. Invocation flags may select a run-specific posture without editing project configuration.
Decision order
- Capability must be healthy and exposed.
- Input must satisfy schema and workspace policy.
- Pre-tool hooks may narrow or block the call.
- Approval policy decides whether operator confirmation is required.
- The result and decision are appended to the session.