liveBook · Manning

Action-by-action control

As explained in Claude at Work

Action-by-action control is a permission technique that lets a user allow, review, or block each external operation individually.

Action-by-action control lets you set trust separately for each operation exposed by a connector: allow it always, require your approval, or block it. section 3.5

Use approval for operations that change data. A connector can read information and perform changes, so review the requested action before authorizing it. section 3.5

Reviewing an email write

Claude drafts an email response and requests approval before the email connector performs the write action. section 3.5

Figure 3.9 An email response drafted by Claude, with approval being requested for the write action on the email connector
CH03 F09 Stoneman

The figure shows an approval request for an email connector write action. section 3.5

How Claude at Work covers it

How deep it goes
Defines allow, ask-first, and block choices, and flags data-changing actions for approval.
Examples
Shows approval requested before an email connector writes a drafted response.
What you can do after
Configure connector permissions by allowing, reviewing, or blocking individual operations.
How it connects
Connects action-level permissions to MCP connectors and OAuth-authenticated services.

This explainer was generated by Manning's knowledge graph from the text of Claude at Work. The section links point to the parts of the book it draws on.