AI サービスが一時的に利用できないため、復旧後に翻訳を補完します。
Claude Code Custom Commands & Skills: Automate Your Workflow India's Most Futuristic AI Conference Is Back – Bigger, Sharper, Bolder d : h : m : s Career GenAI Prompt Engg ChatGPT LLM Langchain RAG AI Agents Machine Learning Deep Learning GenAI Tools LLMOps Python NLP SQL AIML Projects Reading list How to Become a Data Analyst in 2025: A Complete RoadMap A Comprehensive Learning Path to Tableau in 2025 A Comprehensive NLP Learning Path 2025 Learning Path to Become a Data Scientist in 2025 Step-by-Step Roadmap to Become a Data Engineer in 2025 A Comprehensive MLOps Learning Path: 2025 Edition Roadmap to Become an AI Engineer in 2025 A Comprehensive Learning Path to Master Computer Vision in 2025 Best Roadmap to Learn Generative AI in 2025 GenAI Roadmap for Enterprises Large Language Models Demystified: A Beginner’s Roadmap Learning Path to Become a Prompt Engineering Specialist Claude Code Custom Commands & Skills: Automate Your Workflow Mounish V Last Updated : 01 Oct, 2026 5 min read Every dev team has repeatable setup tasks: spinning up a new project, running a deployment checklist, generating a file structure, or typing the same instructions into Claude again and again. Claude Code helps reduce that friction with slash commands, including built-in commands and custom commands you can define for your own workflows. These commands turn repeated instructions into quick triggers, making day-to-day development smoother and more consistent. In this article, we’ll explore Claude Code slash commands, understand how skills work, and build a few practical custom commands along the way. Table of contents What is a Slash command? What are Skills in Claude Code? Pre-Requisites Create a new project Open Claude Code Create another skill Conclusion Frequently Asked Questions What is a Slash command? A slash command is a short text trigger, starting with “/”, that runs a predefined action instantly, instead of you typing out a full instruction every time. You’ve likely seen this pattern before in tools like Slack, Notion, or Discord, where typing “/” brings up a menu of quick actions. The idea is simple: turn something you’d otherwise explain in a sentence into a single word you can type in seconds. Built-In Commands in Claude Code Here are some existing Claude commands and what they do: /model: switches the current model (example: Fable 5.1, Haiku 4.5, Sonnet 5) /rewind: rolls the session back to an earlier point in the conversation /compact: compresses the conversation history to free up context space /clear: wipes the current conversation and starts a fresh session /context: shows a breakdown of how your context window is being used You can find more built-in commands here. What are Skills in Claude Code? Skills are special instructions stored in a markdown files; it can be configured to only be invoked by the user not by Claude. In legacy syntax, this could be achieved with Custom Slash Commands. This special instruction gets added to the context when invoked, and is meant to be used for tasks that are generally not needed for Claude, it can be something rare like deployment. It lives as a markdown file in .claude/skills//SKILL.md, and the sub-folder name becomes the command name: so.claude/skills/deploy/SKILL.md becomes /deploy. Scope of Skills Personal scope: stored in ~/.claude/skills/ (home directory). Works across all your projects, shows as a personal skill in /skills. Project scope: stored in .claude/skills/ in the repository. This can be a common shared skill only if the folder is committed via git. (Else it stays local) Directory scope: stored in /.claude/skills/ inside a subfolder of the repository. Loads for sessions started in or below that folder, so it’s useful for a skill that only applies to one folder of a project. Skill Invocation Demo Let us define two commands that the model can’t invoke on its own, one to add a skeleton structure to a project, and another to create a frontend script in a specific manner. Pre-Requisites Make sure you have VSCode (or any other IDE) and Claude Code CLI installed. Claude Code CLI can be installed using this command: Mac / Linux: curl -fsSL https://claude.ai/install.sh | bash Windows (powershell): irm https://claude.ai/install.ps1 | iex Note: Make sure you authenticate yourself with the API or your account to use Claude. Create a new project Create a new project and make a .claude/skills/ folder at the root of the project Define a skill: create a new_repo/ folder inside it with a SKILL.md file containing the below text: --- name: new_repo description: Add the structure to a new project with backend/, frontend/, and README.md disable-model-invocation: true allowed-tools: Bash(ls *) Bash(mkdir *) --- Create the base project structure in the current working directory: 1. Create a backend/ folder. If it doesn't already contain anything, add a minimal .gitkeep file (or a starter file appropriate to the stack the user mentions, if any) so the folder is tracked. 2. Create a frontend/ folder the same way. 3. Create a README.md at the project root with: - A title based on the project name (ask the user or infer from context/directory name) - A one-paragraph description of the project (infer from the task at hand if not explicitly given) - A "Project Structure" section listing backend/ and frontend/ and their purpose - A "Getting Started" section with placeholder setup instructions for each folder Do not overwrite existing files or folders — check first with ls and only create what's missing. Report back what was created. This command helps me add a skeleton structure to the repository, this can user-level custom command if you create new projects very often (You’ll have to ensure its sitting at the root of your computer) Note: Notice that the disable-model-invocation is True, the description of the skill is not visible to the model and it can’t invoke it until you invoke it using run /new_repo. Open Claude Code Fire up the terminal and type claude Now you can simply type /new_repo and you would be able to execute the command Once, the skill has been executed you would be able to skill that your directory structure has been updated: Create another skill Create a .claude/skills/ folder inside frontend/ and add a create_html/ sub-folder with a SKILL.md file inside it (frontend/.claude/skills/create_html/SKILL.md). This scopes the skill to the frontend directory, and the folder name becomes the command -> /create_html. This ensures the skill operates only in ‘frontend’ directory 2. Add this text in the file: --- name: create_html description: Create a standalone HTML file for a given requirement argument-hint: [requirement] disable-model-invocation: true allowed-tools: Write --- Create a single, standalone HTML file that implements the following requirement: $ARGUMENTS Guidelines: - Output one self-contained .html file — inline ` and `, no build step, no external dependencies unless explicitly requested. - Keep it functional and minimal: only build what the requirement asks for, no extra scaffolding or unrelated features. - Use semantic HTML and accessible markup (labels, alt text, ARIA where relevant). - Make it responsive and readable on both desktop and mobile widths. - Save the file in frontend/index.html, and report the path back to the user Run the command Open claude in that directory by running this terminal command: cd frontend && claude Add the prompt and invoke the command to create the script. Note: Once claude finishes execution, you can open the HTML and see for yourself what Claude has crafted. Conclusion Claude Code Skills help you create commands that give Claude Code users a simple way to save repeatable instructions as markdown files, invoked only by you. Whether scoped personally or per-project, they turn rare, specific tasks like scaffolding a repo or generating a standalone HTML file into short commands, saving time and effort. But the markdown instructions have to be defined clearly for better outputs. Frequently Asked Questions Q1: Where should project-scoped skills be stored? A. In .claude/skills//SKILL.md inside the repository. It doesn’t have to be at the root; Claude Code checks the folder you start in and every parent up to the repo root. Q2: Can Claude invoke a skill on its own? A. Yes, by default. Add disable-model-invocation: true to the frontmatter to make it user-only. Q3: How do I exit the Claude Code CLI? A. Type /exit, or press Ctrl+C twice. Mounish V Passionate about technology and innovation, a graduate of Vellore Institute of Technology. Currently working as a Data Science Trainee, focusing on Data Science. Deeply interested in Deep Learning and Generative AI, eager to explore cutting-edge techniques to solve complex problems and create impactful solutions. BeginnerCommand Line Login to continue reading and enjoy expert-curated content. Free Courses 0 Why AI Needs a Human in the Loop Learn when and how to keep humans in your AI workflows. 4.7 Advanced Strands Agents with MCP Build enterprise-grade agentic AI using Strands SDK and MCP. 4.8 Building AI agents with Amazon Bedrock AgentCore Build and deploy production-ready AI agents using Amazon Bedrock AgentCore. 4.7 Building Multi Agent Systems with Strands Agents Design scalable multi-agent architectures with Strands. 0 Building & Evaluating Agentic AI Systems Master Agentic AI, AI Agents & LangGraph for building autonomous AI agents. Recommended Articles GPT-4 vs. Llama 3.1 – Which Model is Better? Llama-3.1-Storm-8B: The 8B LLM Powerhouse Surpa... A Comprehensive Guide to Building Agentic RAG S... Top 10 Machine Learning Algorithms in 2026 45 Questions to Test a Data Scientist on Basics... 90+ Python Interview Questions and Answers (202... 8 Easy Ways to Access ChatGPT for Free Prompt Engineering: Definition, Examples, Tips ... What is LangChain? What is Retrieval-Augmented Generation (RAG)? Become an Author Share insights, grow your voice, and inspire the data community. Reach a Global Audience Share Your Expertise with the World Build Your Brand & Audience Join a Thriving AI Community Level Up Your AI Game Expand Your Influence in Genrative AI Receive updates on WhatsApp Email address Wrong OTP. Enter the OTP Resend OTP Resend OTP in 45s