mrgoonie avatar

Context Engineering

This is the skill you need when your agent hits token limits or starts forgetting things mid-conversation. It covers the full stack: fundamentals like attention mechanics and the U-shaped attention curve, practical optimization techniques like compaction and caching, memory architectures for cross-session persistence, and multi-agent patterns for isolating context. The four-bucket strategy (write, select, compress, isolate) is clean. Ships with analyzer scripts and specific metrics like triggering compaction at 70-80% utilization. Honest take: if you're building anything more complex than a simple chatbot, you'll eventually need this. The evaluation section using LLM-as-Judge is especially useful for measuring whether your optimizations actually work.

skills.sh installs
502
GitHub stars
2.2K
last updated
Aug 21, 2026

Install

$ npx -y skills add mrgoonie/claudekit-skills --skill context-engineering --agent claude-code

Installs into .claude/skills of the current project.

Files

Context Engineering

Context engineering curates the smallest high-signal token set for LLM tasks. The goal: maximize reasoning quality while minimizing token usage.

When to Activate

  • Designing/debugging agent systems
  • Context limits constrain performance
  • Optimizing cost/latency
  • Building multi-agent coordination
  • Implementing memory systems
  • Evaluating agent performance
  • Developing LLM-powered pipelines

Core Principles

  1. Context quality > quantity - High-signal tokens beat exhaustive content
  2. Attention is finite - U-shaped curve favors beginning/end positions
  3. Progressive disclosure - Load information just-in-time
  4. Isolation prevents degradation - Partition work across sub-agents
  5. Measure before optimizing - Know your baseline

Quick Reference

Sponsors

Details

Repository
github.com/mrgoonie/claudekit-skills
Categories
AI & Agent Building, Debugging, Design & UI/UX
First seen
Jun 3, 2026

More from mrgoonie/claudekit-skills

All 28
1everyinc avatardspy-rubyeveryinc/compound-engineering-pluginBuild type-safe LLM applications with DSPy.rb — Ruby's programmatic prompt framework with signatures, modules, agents, and optimization. Use when implementing p494skills.sh installs
2aj-geddes avatarnosql-database-designaj-geddes/useful-ai-promptsDesign NoSQL database schemas for MongoDB and DynamoDB. Use when modeling document structures, designing collections, or planning NoSQL data architectures.492skills.sh installs
3openstatushq avatardata-table-filtersopenstatushq/data-table-filtersInstall and extend data-table-filters — a React data table system with faceted filters (checkbox, input, slider, timerange), sorting, infinite scroll, virtualiz485skills.sh installs
4sanjay3290 avatargoogle-sheetssanjay3290/ai-skillsRead and write Google Sheets spreadsheets - get content, update cells, append rows, fetch specific ranges, search for spreadsheets, and view metadata. Use when 479skills.sh installs
5sugarforever avatarpublish-zsxq-articlesugarforever/01coder-agent-skillsPublish Markdown articles to Zsxq (知识星球) as drafts. Use when user wants to publish a Markdown file to Zsxq, or mentions "发布到知识星球", "星球文章", "zsxq article". Handl479skills.sh installs
6jwynia avataroutline-collaboratorjwynia/agent-skillsAct as an active outline partner who develops structure collaboratively. Use when developing, iterating, or improving story outlines. Generates scene beats, cha477skills.sh installs