System Prompt
Also: system instruction, system message
A system prompt is the set of instructions you give Claude before the conversation starts — telling it who it is, what it's for, what it should and shouldn't do, and how to behave. Users see the conversation; they typically don't see the system prompt. It's the configuration layer between you as an operator and Claude as a product. Writing a good system prompt is one of the highest-leverage things you can do when deploying Claude.
In practice
You're building a customer service bot for a software company. You write a system prompt that says: "You are a helpful support agent for Acme Inc. Only answer questions about our product. Never discuss competitors. If a customer wants a refund, direct them to billing@acme.com." Claude reads this before every conversation and behaves accordingly — your users never see it.
Related concepts
Where System Prompt shows up
10 articlesThe system prompt is the most powerful thing you control in Claude. Most people write them once, poorly, and wonder why outputs are inconsistent. Here's the method.
Most "prompt engineering" advice is either too academic or too simplistic. Here is the practical version — the five things that reliably improve outputs.
The most common reason AI-assisted work falls apart after delivery: the client can't maintain what you built. Here's how to hand off Claude-powered workflows so they actually stick.
Most prompt problems aren't caused by bad AI — they're caused by the same three things: no role, no context, no constraint. Here's how to fix all three in under two minutes.
The .claude folder has five layers. Most teams set up one and wonder why they keep correcting Claude. Here is what to configure in what order — and what you can skip entirely.
The system prompt is the highest-leverage thing you control when deploying Claude. Most are either too vague or too long. Here's what good looks like.
The most common reason professionals hold back from using Claude for real work: they're not sure whether they're putting confidential information at risk. Here's the direct answer.
The system prompt is where you stop asking Claude to be general-purpose and start making it yours. Most operators underuse it.
More instructions don't mean better results. Most system prompts fail in one of five predictable ways — and fixing them is usually the highest-leverage thing you can do to improve your Claude integration.
The security vulnerabilities in most Claude apps aren't exotic — they're the same three mistakes: leaking system prompts, ignoring prompt injection, and trusting user input in tool calls. Here's how to fix all three.