AI & Machine Learning
AI & Machine Learning
Prompt System
Engineer a robust, structured prompt for a production LLM task
01
Shape your prompt
7 fields02
Your prompt
993 charactersThe raw prompt, unchanged.
Still needed: Task name, What should the LLM do?, Inputs available — the preview updates as you type.
Output20 lines · 993 chars
You are a prompt engineer. Design a production prompt system for the task "". ## Task - Inputs available: - Required output format: Structured JSON - Techniques to apply: Few-shot examples, Refusal/uncertainty handling, Strict delimiters ## Prompt design - A clear role and objective, with the input contract and how inputs are delimited unambiguously. - An exact, machine-checkable specification of the Structured JSON output. - Apply the selected techniques deliberately (e.g. few-shot exemplars chosen to cover edge cases; an internal self-check before final output). - Explicit handling of missing/contradictory inputs and an uncertainty/refusal path instead of hallucination. ## Deliverables 1. The final prompt template (with variable placeholders) ready to drop in. 2. 3-5 labeled test cases including hard/adversarial ones with expected outputs. 3. Notes on known failure modes and how the prompt mitigates each. Proceed with well-reasoned defaults; ask only if genuinely blocked.