Prompt
The text or instructions you give to an AI model or agent. Prompts define the task, tone, context, output format, and the constraints the model should follow. Well-written prompts improve output quality dramatically; many no-code tools let you edit prompts in the UI without coding. For production agents, the prompt is one input among many—the broader practice of curating everything in the context window is called Context Engineering.
Written by Max Zeshut
Founder at Agentmelt · Last updated Sep 9, 2026
Example
A support agent's prompt might read: 'You are a customer support agent for ExampleCo. Answer using only the provided knowledge base articles. If the question is not covered, say so and offer to escalate. Match the customer's tone. Never invent return policies, pricing, or shipping ETAs.' This 4-sentence prompt encodes role, source-of-truth restriction, fallback behavior, tone rule, and explicit no-go zones—far more effective than a generic 'help the customer' instruction.
Frequently asked questions
- What's the difference between a prompt and a system prompt?
- A system prompt is the fixed instruction block the model reads on every request—role, rules, output format. A user prompt is whatever the current user just said. Most production agents combine a stable system prompt with dynamic user input and retrieved context. The system prompt is where you encode behavior; user prompts are what the agent responds to.
- How long should a prompt be?
- As long as needed and not longer. Production system prompts commonly run 500-3,000 tokens. Above ~5,000 tokens, you usually get better results by splitting into Agent Skills that load on demand rather than a monolithic prompt that includes every edge case up front.