System Prompt
Foundational instructions defining an AI's persona, constraints, and baseline behavior.
Definition
A system prompt is a set of foundational instructions provided to a large language model at the beginning of a conversation, establishing its persona, capabilities, constraints, and behavioral guidelines. Unlike user messages, system prompts are typically hidden from end users and serve as the developer's primary mechanism for controlling how the model behaves.
Key characteristics of system prompts include:
-
Persona Definition: System prompts establish who the AI is, including its tone, expertise areas, and communication style. This shapes all subsequent interactions within the session.
-
Behavioral Constraints: They set explicit boundaries on what the model should and should not do, such as refusing to generate harmful content or staying within a specific domain.
-
Output Formatting: Developers use system prompts to specify response formats, including JSON structures, markdown conventions, or length requirements.
-
Context Priming: System prompts can include reference information, examples, or few-shot demonstrations that guide the model toward desired response patterns.
-
Priority Hierarchy: Most model providers treat system prompt instructions with higher priority than user messages, making them the primary layer of control for application developers.