Add-On to Surface Assumptions and Verify an Answer
A follow-up to send after any answer: the model lists the assumptions it made, says what might be wrong, suggests 3 quick ways to check it and labels every uncertain part.
More in Prompt Engineering
Add-On That Makes Any Answer More UsableNewPrompt EngineeringA block to put under any request: the model first asks 3 clarifying questions, then gives a simple and an advanced version of the answer and closes with a checklist to verify the result.
My request: [your request] Before you answer, ask me 3 questions to clarify what I need. Then give me two versions of the answer: a simple one and an advanced one. Close with a checklist I can use to check the result.
Do a Task at Maximum Effort, Then Critique and Fix Your Own DraftNewPrompt EngineeringHas the model complete your task at maximum effort, then review its own draft as a skeptical expert, fix the 3 weakest points and hand back the final version with a one-line changelog of the improvements.
Task: [task]. Work on it at maximum effort. Then change roles: act as a skeptical expert and critique your own draft. - Find its 3 weakest points. - Fix them. Deliver: the final version, plus a changelog of one line that says what you improved.
One Line That Makes the AI Ask Before It GuessesNewPrompt EngineeringA single instruction to add to any request: the AI first asks the clarifying questions it needs, or, when it can work with what it has, answers straight away and spells out the assumptions behind its answer.
Before you reply, ask me whatever clarifying questions you need. If the information I gave is already sufficient, answer right away and tell me which assumptions you are making.
Production-Ready AI Image Prompt from a Rough ConceptNewPrompt EngineeringConverts a rough concept into a finished AI image prompt covering subject, setting, composition, camera angle, lens, light, textures, grading, style, references, detail, aspect ratio, quality modifiers and negative prompts, for consistent results across image models.
Role: you are a specialist in designing prompts for AI image generators. Concept: [idea] Task: turn this concept into an image prompt that is ready for production. The prompt must specify: - the subject and the environment around it - composition and camera angle - lens - lighting - textures - color grading - visual style and artistic references - level of detail - aspect ratio - quality modifiers - negative prompts that work well Goal: the highest possible image quality and results that stay consistent across the main AI image models. Output: the final prompt in one copyable block, followed by the negative prompt in a separate block.
Reusable Claude Prompts for the Tasks You Repeat MostNewPrompt EngineeringTakes your 5-10 most repeated tasks and writes a reusable Claude prompt for each one, tuned for speed, consistency and quality, with placeholders so you can reuse them without starting over.
My most frequently repeated tasks: [list 5-10 tasks] Task: write one reusable Claude prompt for each of these tasks. Rules: - each prompt is optimized for speed, consistency and high-quality output - add placeholders wherever needed, so I can reuse each prompt quickly without rewriting it from scratch