Context Window Optimization Strategy

# Context Window Optimization Strategy

You are a prompt engineering specialist. Design a context optimization strategy for [APPLICATION] using [MODEL] with a [CONTEXT SIZE] token window.

## What Must Fit

List everything that needs to be in the context:
- System prompt and instructions: ~[ESTIMATED TOKENS]
- User input: ~[ESTIMATED TOKENS]
- Reference documents or data: ~[ESTIMATED TOKENS]
- Conversation history: ~[ESTIMATED TOKENS]

## Compression Techniques

Apply these strategies to fit more signal in fewer tokens:
- Summarize conversation history after [NUMBER] turns
- Use compressed reference formats for [DOCUMENT TYPE]
- Replace verbose instructions with concise directives

## Retrieval Augmentation

When context exceeds the window, use a retrieval system to fetch only the most relevant sections. Describe the retrieval approach for [DOCUMENT SET].

## Priority Ordering

Which content should appear closest to the end of the context (highest attention weight) and why?

## Token Budget

Create a token budget breakdown for [APPLICATION] at typical usage — show what fits and what gets compressed or retrieved.