What Are Agent Skills and How To Use Them

Agent Skills are an open standard for packaging and reusing workflows, domain expertise, and capabilities into portable folders, allowing AI agents to discover and utilize them efficiently. These skills, originally devised by Anthropic, are now an open standard adopted by over 26 platforms, including OpenAI Codex and GitHub Copilot. They address common issues like context window overload, lack of portability, and inconsistency in output by loading only essential information at startup and progressively disclosing more as needed. A skill comprises a folder with a required SKILL.md file containing YAML frontmatter and markdown instructions, and may include optional scripts, references, and assets. This structure allows for repeatable and predictable workflows across different platforms, promoting consistency and efficiency. Skills can be combined with other agent components like MCP servers and subagents to create complex workflows, ensuring that the agent knows how to utilize tools and data effectively. Anthropic's open standard encourages cross-platform portability and community collaboration, with skills available from partners like Atlassian and Zapier, enhancing the versatility and integration of AI systems.