Skip to content
Get packs

Agent Skills for Context Engineering

A collection of reusable instruction packs (skills) that teach AI coding agents how to manage their context wi

GitHub Download v2.3.0
Stars
18.3k
Forks
1.5k
Licence
MIT
Language
Python
Last update
25 days ago
Type
Curated list

What it is

This repository is a set of Agent Skills, written as markdown guides, covering context engineering: deciding what information goes into an AI model's limited context window and how to keep it useful over long sessions. It covers topics such as compression, memory, tool design, multi-agent patterns, and evaluation. It is designed to be loaded by agent platforms like Claude Code, which activate relevant skills based on the task.

  • Load the context-degradation skill to help diagnose lost-in-the-middle behavior in an agent
  • Use the context-compression skill when planning how to shorten long tool outputs or conversation history
  • Consult the multi-agent-patterns skill when deciding whether parallel agents are justified

Great for

  • Developers building or debugging LLM agent systems that run long sessions
  • Engineers who want a structured reading path on context management and multi-agent design
  • Claude Code users who want these skills available as a plugin

Not ideal for

  • People looking for a runnable library or application, since the README describes guides and pseudocode rather than a finished package
  • Beginners with no experience of LLMs or agents, since many skills assume familiarity with these concepts
Step-by-step guide: how to use Agent Skills for Context EngineeringInstall, first run, examples, tips and FAQs

How to get it

  1. 1

    See the README

    You'll need: Claude Code or another agent platform that supports skills

    Register the marketplace in Claude Code

    /plugin marketplace add muratcankoylan/Agent-Skills-for-Context-Engineering

    Install the plugin directly

    /plugin install context-engineering@context-engineering-marketplace
  2. 2

    Run it

    • Run the marketplace add command inside Claude Code
    • Run the plugin install command, or use the browse-and-install menu
    • Start a task; skills activate automatically based on context
  3. 3

    Get the most out of it

    • Read the Skill Activation Scenarios table to see which skill fits which situation
    • Start with the Foundational Skills (context-fundamentals, context-degradation) before the advanced ones
    • Use advanced-evaluation when you need to compare agent outputs with an LLM judge

Downloads

Source code (.zip)main branch · straight from GitHubRelease v2.3.04 files incl. researcher-os-pipeline-v2-3.svg

Strengths

  • Covers a broad set of context engineering topics in one place
  • Designed to be platform-agnostic according to the README, not tied to one vendor
  • Has a clear skill-activation table mapping tasks to skills

Trade-offs

  • The README describes its scripts as Python pseudocode, so they are not a ready-to-use codebase
  • Several skills are marked NEW and reference specific research or prompts that this entry cannot independently verify from the README

Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.