System message of an agent defining its role, objectives, tone and boundaries; it also steers, though not deterministically, when tools get called.
Also called system message.
Read more: Microsoft Learn
In the Ultra Transcenders books
Each book explains Agent instructions in context, with comparison tables and the common traps.
Terms in this definition
- Agent
A specialised form of Microsoft Copilot set up for one particular job, pairing instructions with knowledge and skills. You can create one in Copilot Studio, SharePoint or Agent Builder, and administrators control them from the Microsoft 365 admin center.
- Role
How an actor normally or expectedly behaves, or the part a person takes in a process. A single actor may hold more than one role.
- Get
Key Vault permission on secrets that allows a single secret to be read; App Service Key Vault references need nothing beyond it.
Related terms
- Chat playground
Lets you try a deployed chat model in Foundry without writing code, adjusting the system message, temperature, past messages included and maximum response length.
- Metaprompt and grounding mitigation layer
Layer of generative AI mitigation at the application level, in which grounding data and the system message guide the model away from output that is ungrounded or harmful.
- Safety system message
A system message that spells out boundaries and how to refuse in order to reduce harm. It is just one layer in a safety approach rather than a full control on its own.
- User message
The chat role for whatever the end user supplies, such as a question or an image, in contrast to the persistent instructions held in the system message.
- User prompt attack
Also called a jailbreak, this is a harmful prompt typed by the user to override the system message or safety training. Prompt Shields for user prompts flags it; instructions concealed in documents are a different threat it misses.