Review System Prompt

A system prompt is a set of higher-priority instructions supplied by a model provider or application developer to shape an AI assistant's role, behavior, boundaries, and response style.

The user usually writes the visible request, while the application may also send background instructions that establish how the assistant should respond. Exact instruction priority and visibility depend on the system.

System prompts can explain consistent product behavior, but they are only one control layer and can interact with tools, policies, retrieval, user input, and model limitations.

A support assistant receives background instructions to answer only from approved documentation and to hand uncertain cases to a person.

A system prompt is not an absolute security boundary. It guides behavior but still requires testing, access controls, and application-level safeguards.

The application sets the role before the user makes the request.

* [[prompt|Prompt]] * [[prompt-injection|Prompt Injection]] * [[guardrails|Guardrails]] * [[policy-leakage|Policy Leakage]]

* [NIST AI 100-2 E2025, Adversarial Machine Learning](https://doi.org/10.6028/NIST.AI.100-2e2025) — NIST terminology for system prompts and prompt-based attacks.

* Contrast intended instructions with [[prompt-injection|Prompt Injection]].