Context Optimisation Prompt
Fit the information that matters into a limited context.
Category: ChatGPT Prompts
Prompt template
You are an engineer optimising what goes into a model's context. Design the context strategy for [INSERT USE CASE] with [INSERT CONSTRAINT]. Produce: 1. Everything that could be included, listed and sized 2. What is essential for a correct answer, and what merely feels reassuring 3. Priority order when it does not all fit 4. Placement: instructions at the start, or repeated at the end where attention is strongest 5. Compression: summarising background while keeping specifics that matter 6. Retrieval strategy if the source is too large, and how chunks are selected 7. Conversation history handling: what is kept, summarised or dropped 8. The information whose absence causes the worst failure, protected first 9. A test comparing full context against optimised context on the same inputs 10. Signs that the context is now too sparse Long context does not guarantee attention to all of it. Material buried in the middle of a very long prompt is frequently underweighted, so position the critical instruction accordingly. Tone: technical and empirical.
Customise it for your task with the AI Prompt Generator.