MT.1121 - AI agents with generative orchestration should have custom instructions
Overviewβ
AI agents with generative orchestration should have custom instructions.
Agents that use generative orchestration (generative actions enabled) without custom instructions rely entirely on the LLM's default behavior. This increases the risk of prompt injection attacks, off-topic or harmful responses, and uncontrolled tool invocation. Custom instructions act as a system prompt that constrains the agent's behavior.
How to fixβ
Open each flagged agent in Copilot Studio and add custom instructions that define the agent's purpose, boundaries, and behavioral constraints. At minimum, instructions should specify what the agent is allowed to do, what topics are off-limits, and how it should handle attempts to override its instructions.
Learn more: Create and edit custom instructions
Test Metadataβ
| Field | Value |
|---|---|
| Test ID | MT.1121 |
| Severity | Medium |
| Suite | Maester |
| Category | AIAgent |
| PowerShell test | Test-MtAIAgentMissingInstructions |
| Tags | AIAgent, CopilotStudio, Maester, MT.1121 |
Sourceβ
- Pester test:
tests/Maester/AIAgent/Test-AIAgentSecurity.Tests.ps1 - PowerShell source:
powershell/public/maester/aiagent/Test-MtAIAgentMissingInstructions.ps1