What problem does it solve? LLM agents degrade unpredictably as context grows: instructions get lost in the middle of long prompts, tool outputs silently consume most of the window, and nominal token limits mislead capacity planning. This Skill provides the foundational mental models and concrete thresholds needed to design, debug, and optimize agent context. ## Core Features & Use Cases - Context anatomy guidance: Covers system prompts, tool definitions, retrieved documents, message history, and tool outputs with actionable placement and sizing rules. - Attention and budgeting mechanics: Explains the U-shaped attention curve, the 60-70% effective-capacity rule, and compaction triggers at 70-80% utilization. - Progressive disclosure patterns: Three-level loading strategy (skill selection, document loading, tool result retention) with a Python context_manager.py utility for token estimation, truncation, validation, and lazy file loading. - Use Case: When an agent's reasoning quality collapses after 20-30 tool calls, use this Skill to diagnose message-history bloat and implement observation masking plus compaction triggers. ## Quick Start Ask the agent to explain why my agent loses track of instructions in long conversations and how to fix its context budget.
