How to Approve Agent Actions and Manage Tool Permissions in MorningAI

MorningAI now gives you direct control over sensitive agent actions through in-conversation approvals and connector-level permissions.

Product: Agents

Why this matters

You can let agents move quickly without giving unrestricted access. Approval prompts let you review actions in context, while tool permissions let your team pre-define what is allowed per connector.

Step-by-step setup

1. Open the permissions panel from your connector settings.
2. Choose a connector, then review the available actions by category.
3. Allow only the actions your team needs for day-to-day workflows.
4. In chat, review each approval prompt when the agent requests a sensitive action.
5. Approve to proceed, or decline to keep control of the final decision.

Common use cases

• Require approval before drafting or sending responses tied to customer accounts.
• Allow read access for research tasks while restricting write actions to managers.
• Set stricter permissions for shared or high-risk connectors.

Pro tips

• Review permissions whenever you add a new connector, not only during initial setup.
• Start restrictive, then expand access as you confirm workflows and team ownership.

FAQ

Do approvals replace permissions? No. Permissions define what the agent may attempt, and approvals confirm high-sensitivity actions at runtime.

Can I change permissions later? Yes. You can update connector action permissions at any time as policies change.

Next steps

If you have not connected Google tools yet, set those up first so your permission rules apply to live agent workflows.