Agent Skills
- Source
- anthropic.com
- Author
- Anthropic
- Date
Context
Anthropic's Agent Skills design packages task knowledge in folders that an reads as needed. Initially, the agent receives each 's name and description. When one matches the task, it reads the instructions, then any relevant supporting files. A document-processing skill can therefore keep general guidance separate from detailed form-filling instructions. Bundled scripts can perform operations without putting their entire contents into the conversation. This gives teams a way to reuse procedures while limiting how much reference material occupies the model's working .
- AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
- agent skill — A reusable instruction file that teaches an agent how to do one job well — the procedure, the tools, and what counts as done.
- context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
“Building a skill for an agent is like putting together an onboarding guide for a new hire.”
“Progressive disclosure is the core design principle that makes Agent Skills flexible and scalable.”
“Skills provide Claude with new capabilities through instructions and code.”
“Skills are a simple concept with a correspondingly simple format.”
Checking sign-in…
Loading comments…