System Instructions (System Prompt and Custom Instructions)
The primary hidden directive from the developer or user (System Message). It establishes fundamental behavioral frameworks, roles, communication styles, and prohibited topics that the model retains throughout the session.
1. Concept Overview & Systemic Problem
When you simply open a chat and ask a question, the model responds in its standard averaged style: with lengthy polite introductions, template advice, and a desire to please everyone simultaneously.
System Instructions (System Prompt / Custom Instructions) are a way to "reprogram" the character and behavior of artificial intelligence to suit your needs. This hidden root rule is invisibly added at the very beginning of each of your queries.
For beginners, system instructions are the secret button that permanently frees you from the need to repeat the same introductory requirements each time.
2. Architectural Taxonomy & Mental Model
Every conversation under the hood resembles a structured play with three actors:
┌─────────────────────────────────────────────────────────────┐
│ THREE LEVELS OF DIALOGUE WITH AI │
├─────────────────────────────────────────────────────────────┤
│ 👑 1. SYSTEM (Rules of Engagement): │
│ "You are a concise senior DevOps engineer. │
│ Respond only with code and bash commands without │
│ unnecessary greetings or apologies." │
├─────────────────────────────────────────────────────────────┤
│ 👤 2. USER (User): │
│ "How do I check free disk space in Linux?" │
├─────────────────────────────────────────────────────────────┤
│ 🤖 3. ASSISTANT (Artificial Intelligence): │
│ `df -h` │
└─────────────────────────────────────────────────────────────┘
3. Golden Template for Custom Instructions for Your Account
Copy this template into the Custom Instructions section in your ChatGPT or Claude settings to instantly enhance your interaction quality:
Field 1: What should the assistant know about you? "I value my time and high technical accuracy. My knowledge level is basic engineering. I use macOS and work with web technologies."
Field 2: How should the assistant respond? "1. Respond in Ukrainian. 2. Be concise: get straight to the point without introductions like 'Of course, I will help!'. 3. If there is a non-obvious pitfall or risk in my solution, directly warn me about it. 4. If you are not 100% sure about a fact, honestly say: I don't know, instead of making something up."
4. Production Engineering Scenarios
Setting system rules once gives you an assistant that:
- Never starts a response with boring compliments.
- Immediately provides information in a format convenient for you (table, code, or bullet points).
- Always remembers the language you prefer to communicate in.
01. Enhancing Team Collaboration
02. Streamlining Code Reviews
03. Automating Routine Queries
5. Pitfalls, Common Mistakes & Security
Be cautious of over-relying on system instructions without regular updates, as they may become outdated with evolving project requirements. Always validate the model's responses against current best practices to avoid potential hallucinations or inaccuracies.
FAQ: System Instructions (System Prompt and Custom Instructions)
Related terms
System Prompt (System Instructions & Metaprompting)
The primary metacontext block of instructions passed at the zero position of the context window, defining the agent's role, safety rules, available tools, and behavioral boundaries.
Role Prompting: How Role Assignment Radically Changes Response Quality
A technique for formulating prompts by assigning the model a specific role, profession, or psychological persona (e.g., 'Act as a Senior DevOps', 'Imagine you are a pediatrician'). This activates specific layers of knowledge and professional vocabulary.
Delimiters and XML Tags (Structuring Prompts)
A technique for formatting complex prompts using XML tags (<context>, <rules>) and triple quotes (\"\"\"). It helps the model clearly distinguish where your rules end and where the text for processing begins.