{"slug": "are-you-missing-out-on-agent-skills-here-s-how-they-work", "title": "Are You Missing Out on Agent Skills? Here's How They Work", "summary": "Rijul, the developer behind git-lrc, explains the concept of 'agent skills' as a way to give AI agents procedural knowledge through a SKILL.md file. Skills use progressive disclosure to load only necessary information, and they differ from MCP, RAG, and fine-tuning. The post also warns about security risks from malicious skills.", "body_md": "*Hello, I'm Rijul. I'm building git-lrc, a micro AI code reviewer that runs on every commit. It's free and source-available on GitHub. Star git-lrc to help more developers discover the project. Do give it a try and share your feedback*\n\nDo you keep hearing terms like “agent skills” being thrown around without really understanding what they mean?\n\nAnd is the FOMO starting to get to you?\n\nLet’s understand what agent skills are in this article.\n\nA skill is a way to give an AI agent **procedural knowledge**.\n\nAt its core, a skill is simply a `SKILL.md`\n\nfile.\n\nBut what exactly does this file contain?\n\nAt the top of the file, we have the YAML frontmatter.\n\nAt a minimum, it contains two things:\n\nThis identifies the skill.\n\nThis tells the agent:\n\nThe description essentially acts as the trigger condition for the skill.\n\nBelow the frontmatter, we have the body of the file.\n\nThis contains the actual instructions.\n\nIt explains what the agent needs to know to perform the task.\n\nFor example:\n\nSkills can also contain optional directories.\n\nThis can contain executable code that the agent can run when needed.\n\nThis directory can contain additional documents that the agent can refer to.\n\nThis can contain static resources that may be needed while performing the task.\n\nAn agent can have a large number of skills.\n\nBut this raises an obvious question:\n\n**If an agent has hundreds of skills, does loading all of them consume a huge amount of context?**\n\nThis is where a concept called **progressive disclosure** comes in.\n\nThe information is loaded in three stages.\n\nWhen the agent starts, it loads only the name and description of each skill.\n\nThis allows the agent to understand what skills are available without loading all of their instructions into the context.\n\nWhen the description of a skill matches the task, the agent loads the full instructions from that skill into its context.\n\nScripts, references, and assets are loaded or used only when they are needed.\n\nThis prevents the agent from loading everything at once.\n\nThere are a few concepts that may sound similar.\n\nMCP gives agents the ability to interact with external tools and APIs.\n\nIt is mainly about **what the agent can reach**.\n\nRAG retrieves relevant information from a knowledge base.\n\nIt is mainly about giving the agent **reference material and factual knowledge**.\n\nFine-tuning changes the model's weights.\n\nThe knowledge or behavior is effectively baked into the model itself.\n\nSkills are different.\n\nThey provide the agent with **procedural knowledge**.\n\nThey tell the agent:\n\nA skill can also use other technologies, such as MCP.\n\nFor example, a skill could provide the instructions and judgment for **when to call a particular MCP tool**.\n\nThis format is based on the open standard from Agent Skills.\n\nIt has also been adopted by AI platforms such as Claude and Codex.\n\nThis makes skills a useful way to package and share procedural knowledge across different AI tools.\n\nThere is one important thing to keep in mind.\n\nSkills can execute scripts.\n\nThat means installing a skill is not always as simple as downloading a text file.\n\nA malicious skill could potentially introduce risks such as:\n\nSo you should treat skill installation similarly to installing any other software.\n\nOnly install skills from sources you trust, and review what they contain before giving them access to your AI tools.\n\nSo that's it about this article, see you on the next one!\n\nAI agents write code fast. They also silently remove logic, change behavior, and introduce bugs -- without telling you. You often find out in production.\n\n[git-lrc](https://github.com/HexmosTech/git-lrc) fixes this. It hooks into git commit and reviews every diff before it lands. 60-second setup. Completely free.\n\nAny feedback or contributors are welcome! It's online, source-available, and ready for anyone to use.\n\nGive it a ⭐ [star on Github](https://github.com/HexmosTech/git-lrc)", "url": "https://wpnews.pro/news/are-you-missing-out-on-agent-skills-here-s-how-they-work", "canonical_source": "https://dev.to/rijultp/are-you-missing-out-on-agent-skills-heres-how-they-work-4dk5", "published_at": "2026-07-17 20:32:30+00:00", "updated_at": "2026-07-17 20:59:58.493211+00:00", "lang": "en", "topics": ["ai-agents", "developer-tools"], "entities": ["Rijul", "git-lrc", "Claude", "Codex", "Agent Skills", "GitHub"], "alternates": {"html": "https://wpnews.pro/news/are-you-missing-out-on-agent-skills-here-s-how-they-work", "markdown": "https://wpnews.pro/news/are-you-missing-out-on-agent-skills-here-s-how-they-work.md", "text": "https://wpnews.pro/news/are-you-missing-out-on-agent-skills-here-s-how-they-work.txt", "jsonld": "https://wpnews.pro/news/are-you-missing-out-on-agent-skills-here-s-how-they-work.jsonld"}}