context▌

When agent sessions generate millions of tokens of conversation history, compression becomes mandatory. The naive approach is aggressive compression to minimize tokens per request. The correct optimization target is tokens per task: total tokens consumed to complete a task, including re-fetching costs when compression loses critical information.

golang-context

Strategies for managing LLM context windows through summarization, trimming, routing, and prioritization. \n \n Covers six core capabilities: context engineering, summarization, trimming, routing, token counting, and prioritization to prevent token limits and context rot \n Implements tiered context strategies that adapt based on context size, and serial position optimization to place critical information at start and end of context \n Avoids common anti-patterns including naive truncation, igno

$23

Context optimization extends the effective capacity of limited context windows through strategic compression, masking, caching, and partitioning. The goal is not to magically increase context windows but to make better use of available capacity. Effective optimization can double or triple effective context capacity without requiring larger models or longer contexts.

Context optimization extends the effective capacity of limited context windows through strategic compression, masking, caching, and partitioning. The goal is not to magically increase context windows but to make better use of available capacity. Effective optimization can double or triple effective context capacity without requiring larger models or longer contexts.

Tell the agent to zoom out and give broader context or a higher-level perspective.

Diagnose whether your AI workflows are context stuffing or context engineering, then implement structured practices to improve output quality. \n \n Distinguishes between context stuffing (volume-based) and context engineering (structure-based), with five diagnostic questions to identify Context Hoarding Disorder \n Guides two-layer memory architecture: short-term conversational memory plus long-term persistent memory via vector databases for semantic retrieval \n Implements the Research→Plan→Re

Analyze codebases and map task-relevant files before implementing changes. \n \n Automatically searches for related files, dependencies, tests, and similar code patterns to establish full context \n Generates structured markdown output with files to modify, dependency relationships, test coverage, and reference patterns \n Includes risk assessment checklist for breaking changes, migrations, and configuration updates \n Designed as a pre-implementation step to prevent missed dependencies and unin

Ask an AI assistant what files it needs before answering your question. \n \n Prompts the assistant to identify required and optional files for context before attempting to answer \n Structures output into four categories: must-see files, helpful files, already-seen files, and remaining uncertainties \n Helps developers avoid incomplete answers by ensuring the assistant has examined relevant code upfront \n Reduces back-and-forth by clarifying dependencies and gaps in context before the actual q