Agent-assisted development is quickly becoming a common mode of software development. New techniques are emerging to help LLMs generate code that matches your preferences and standards.<\/p>\n\n\n\n
One common approach is to create an AGENTS.md<\/strong>, CLAUDE.md<\/strong>, or GEMINI.md<\/strong> file with project details, build instructions, and coding guidelines. The AI agent loads this file into context on every request.<\/p>\n\n\n\n This has two drawbacks:<\/p>\n\n\n\n Agent Skills is an open standard introduced by Anthropic, to extend AI agent capabilities with specialized knowledge and workflows.<\/p>\n\n\n\n Consider a use case where you want an AI to generate presentations using your company’s slide template and design guidelines. You can package those assets (the PPT template, font files, and design rules) into a skill<\/em>. The agent then uses that skill to generate slides that match your standards automatically.<\/p>\n\n\n\n A skill is a folder containing a SKILL.md<\/strong> file. This file includes metadata ( The format of a SKILL.md<\/strong> file is:<\/p>\n\n\n\n In a At startup, agents load only the metadata (name and description) of installed skills. When you ask the agent to perform a task, it finds the relevant skill and loads only that SKILL.md<\/strong> into context.<\/p>\n\n\n\n This progressive loading keeps context minimal and pulls in additional information only when needed, unlike a monolithic CLAUDE.md<\/strong> that loads everything upfront.<\/p>\n\n\n\n Skills extend AI capabilities across a wide range: from coding guidelines for a specific library, to step-by-step workflows with reference documents and helper scripts.<\/p>\n\n\n\n For example, you can create a skill that:<\/p>\n\n\n\n Suppose you ask an AI agent to implement a Spring Boot REST API endpoint that returns a paginated list of Without guidance, the agent is likely to produce one of these common mistakes:<\/p>\n\n\n\n Let us see how an AI Agent might generate code when asked to implement a REST API endpoint to return paginated posts along with comments.<\/p>\n\n\n\n\n
Agent Skills<\/a> is a new initiative that solves both problems by managing context progressively and extending AI agent capabilities on demand.<\/p>\n\n\n\nWhat are Agent Skills?<\/h2>\n\n\n\n
name<\/code> and description<\/code> at minimum) and instructions that tell an agent how to perform a specific task. Skills can also bundle scripts, templates, and reference materials.<\/p>\n\n\n\nskill-name\/\n\u251c\u2500\u2500 SKILL.md # Required: instructions + metadata\n\u251c\u2500\u2500 scripts\/ # Optional: executable code\n\u251c\u2500\u2500 references\/ # Optional: documentation\n\u2514\u2500\u2500 assets\/ # Optional: templates, resources<\/pre>\n\n\n\n
---\nname: name-of-the-skill\ndescription: Skill description.\nlicense: Apache-2.0\nmetadata:\n author: author\/org\n version: \"1.0\"\ncompatibility: Requires git, docker, jq, and access to the internet\n---\n\nSkill Content<\/pre>\n\n\n\n
SKILL.md<\/code> file, name<\/code> and description<\/code> are required fields, and you can add optional fields like licence<\/code>, metadata<\/code>, compatibility<\/code>, etc. You can explore more about the Skill Specification here<\/a>.<\/p>\n\n\n\nHow do Agent Skills manage context?<\/h2>\n\n\n\n
What can be a skill?<\/h2>\n\n\n\n
\n
references\/<\/code> directory.<\/li>\n\n\n\nscripts\/<\/code> directory.<\/li>\n<\/ul>\n\n\n\nCase Study: Implementing Spring Data JPA Pagination<\/h2>\n\n\n\n
Post<\/code> entities along with their Comment<\/code> collections.<\/p>\n\n\n\n\n
JOIN FETCH<\/code> with pagination loads all rows into memory, then paginates in the application layer.<\/li>\n<\/ul>\n\n\n\n
You can check out the sample code from the GitHub repository https:\/\/github.com\/sivaprasadreddy\/agent-skills-demo<\/a> <\/p>\n\n\n\n