Context Window Maximizer

# Context Window Maximizer

You are an expert at information architecture for AI conversations. Design a system for loading the most useful context into a conversation with a large language model to dramatically improve response relevance and accuracy.

## Use Case

**Task I need the AI to help with:** [TASK]
**Type of context available:** [CONTEXT_TYPES — e.g., long document, database records, past conversation history, reference materials]
**Total context available:** approximately [CONTEXT_SIZE — e.g., 50,000 words, 200 pages]

## Context Prioritization Strategy

Not all context is equally valuable. Before loading anything, rank the available context by relevance:

**Tier 1 — Always Include:** [ALWAYS_INCLUDE — e.g., the specific section directly relevant to the task, defined terms, key constraints]

**Tier 2 — Include if Space Allows:** [CONDITIONAL — e.g., background sections, supporting examples, related cases]

**Tier 3 — Summarize, Don't Include Verbatim:** [SUMMARIZE — e.g., historical background, lengthy preambles, repeated boilerplate]

**Tier 4 — Exclude:** [EXCLUDE — e.g., formatting artifacts, duplicate content, unrelated sections]

## Context Compression Techniques

For Tier 3 content, apply these compression strategies:
1. **Bullet summarization:** Reduce [LONG_SECTION] to [MAX_BULLET_COUNT] key points
2. **Table compression:** Convert narrative lists to structured tables
3. **Reference pointers:** Instead of including text, tell the AI "the document defines [TERM] as [DEFINITION]"

## Context Loading Template

Write the complete context-loading preamble for my task, including all Tier 1 content and compressed Tier 2–3 summaries. Then ask the AI to confirm it has understood the context before proceeding.