What is an agent skill? How to protect your agent's context window.

What’s the difference between a long-running agent that can seamlessly execute complex tasks, and one that constantly get stuck? The context. A bad agent is usually the outcome of high volume of information being crammed into a context window all at once. Skills are how LangChain's Deep Agents harness fixes that. A Skill is a folder with instructions, scripts, references, and assets for one task, but only its name and description load at startup. It’s like reading the back of a book instead of the whole thing. In this Short, learn how: Skills keep an agent's system prompt from turning into a wall of formatting rules, tone guidelines, and one-off instructions The agent picks the right skill and pulls it in only when the task calls for it Context disappears once a task is completed. Go deeper for free — Foundation: Introduction to Deep Agents on LangChain Academy: → Course: https://academy.langchain.com/courses/foundation-introduction-to-deepagents → Deep Agents on GitHub (MIT licensed): https://github.com/langchain-ai/deepagents → LangGraph: https://www.langchain.com/langgraph #AIAgents #LangChain #DeepAgents