# Agent permissions: decide what each agent step may do — Orlo

Markdown version of https://useorlo.com/agents.html, generated 2026-10-13 from the page itself. Illustrations and interactive examples are left out.

**Decide what your agents may do.**

Set tool permissions, require approval for selected actions, and limit how far a run can go. Orlo records the steps you send it and the decisions it returns.

## Set permissions for each tool.

In this workspace, looking up an order is allowed, issuing a refund needs approval, and closing an account is blocked.

## Approve or reject a proposed action.

Orlo checks the proposed action. Your runtime enforces the decision. A person steps in where your rules require one.

The runtime remains responsible for executing tools and enforcing the decision.

Explore an example approval decision.

### Set limits for the run.

Set limits on steps, tool calls, and write actions. Give your runtime a clear point to stop.

### Require approval before execution.

Require approval for a proposed action. Your runtime waits for that decision before executing it.

### Review the rule behind a permission decision.

Review the proposed action, the rule it matched, and the approval decision in the run’s record.

## Check live answers against your task’s rules.

## More

Orlo is a fineSample product. Documentation: https://docs.useorlo.com/

Other pages as Markdown: [Overview](https://useorlo.com/index.md), [Evaluation](https://useorlo.com/evaluation.md), [In production](https://useorlo.com/deployment.md). Index: https://useorlo.com/llms.txt