The primary background instruction set passed to a language model that sets its persona, behavioral constraints, tone, and operational boundaries before user prompts are processed.

A core MLOps building block. It drapes writing hidden background text instructions in authoritative systemic command language.

• Authoritative Command Posture: Framing hidden background instructions as unbreakable systemic law.
• The System Prompt Override: Claiming absolute behavioral control while user prompt injection tricks frequently bypass system prompt rules.
• House-Cleaning Reality Check: Like writing rules on a bulletin board and calling it a "system sanitation prompt."

The Terminator (1980s) scene where the T-800's red heads-up display displays scrolling system directives, setting target priorities and behavioral rules.

“Our system prompt instructs the model to refuse requests for confidential employee information.”

“We wrote background instruction rules telling our search tool to maintain a professional tone and refuse questions about confidential salary data.”