Pause an agent tool call and review its exact action and parameters.
Workflow & automation · Documented · Documentation checked 2026-10-04
n8n Cloud or a maintained self-hosted instance; models and connected apps are external services.
Instance version, edition, hosting, integration credentials, and model provider access affect capabilities.
The worked example is illustrative and has not been tested in a signed-in product account.
What it does and how it connects
Pause an agent tool call and review its exact action and parameters.
Human review pauses an agent before a selected tool runs. The reviewer sees the tool and proposed parameters, then approves or denies. This is stronger than merely asking the model to be careful because the workflow actually waits at the action boundary.
Walkthrough
Use the product named in the access panel. The steps below combine documented controls with the editorial example.
- Open the AI Agent Tools connector / tools panel in a supported instance.
- Add a human-review step using an available approval channel and its credentials.
- Connect the sensitive tool under the review step and configure its operation.
- Test approval and denial with a sandbox action before allowing real destinations.
Worked example
A draft invitation can be prepared by AI, but sending should require review of the exact recipient, subject, and body. Use a test inbox first. A denial should cancel the tool call and leave a visible record of the decision.
Propose an invitation draft for the specified workshop. Use the authorized send tool only through the configured human-review gate. Show the exact recipient and content. If the request is denied or the recipient is ambiguous, stop; do not use another channel to send it.
Inspect and verify the result
Observe that the workflow pauses before the action. Test denial and confirm no message was sent. Test approval in a sandbox and compare actual tool parameters with what the reviewer saw.
Limits and troubleshooting
An approval of one tool call is not blanket approval of every future call. Verify reviewer identity, message delivery, and behavior when nobody responds. Prompts alone cannot prevent a different ungated tool from doing the same action.
Data, permissions, and actions
The approval channel receives proposed action details and can disclose personal information. Restrict the reviewer audience and tool credentials. Keep every equivalent consequential operation behind a gate or unavailable.
Related reading
Evaluation, errors, and execution history
Test output quality and operational failures as separate parts of reliability.
A fixed classification-to-review workflow
Transform a fictional workshop request into a validated private review record.
A FAQ tool agent with action approval
Answer approved workshop questions and pause before any sensitive tool action.
Permissions and meaningful approvals
Decide what a system may read, what it may change, and when someone needs to authorize a concrete action.
Sources and verification
Documentation checked 2026-10-04. Examples are illustrative unless described otherwise.