# Limit an agent’s authority

Taiga Learning · Worksheet
https://taiga.training/en/lessons/agent-permissions/

Use fictional or approved information. Do not put secrets in this worksheet.

## Learning objectives
- Express a permission as an action, resource, and condition.
- Separate task approval from authorization to execute.
- Test that a forbidden action is denied.

## Exercise
Define permissions for an agent that fixes a report filter. List three allowed actions and three denied actions. Include repository, branch, environment, and expiry. Describe how to test each denial without changing production.

## Your response
- Scenario and scope:
- Assumptions and open questions:
- Proposed answer or decision, with reasons:

## Verify your response
| Claim or criterion | Evidence or test | Result or gap | Owner |
| --- | --- | --- | --- |
| | | | |
| | | | |
| | | | |

## Next action
- Action, owner, and date:
- When will you review this response?

## Principle to retain
An agent’s instructions describe intended behavior. The tool and identity system must enforce its actual authority.

## Sources
- [AWS IAM: security best practices](https://docs.aws.amazon.com/IAM/latest/UserGuide/best-practices.html)
- [OWASP: AI Agent Security](https://cheatsheetseries.owasp.org/cheatsheets/AI_Agent_Security_Cheat_Sheet.html)

This worksheet supports learning. Completing it does not itself authorize a production change.