tutorial — команда для Claude Code: что делает и как установить
Команда tutorial для Claude Code знакомит с агентами, командами, хуками и навыками плагина на практических примерах, не изменяя файлы проекта.
MIT ★ 1 424 обновлён 2026-10-06
.claude/commands/tutorial.mdЧто делает tutorial
Слэш-команда tutorial запускает пошаговое знакомство с окружением плагина в режиме чтения. Сначала она ищет маркеры проекта через Glob (package.json, pyproject.toml, Cargo.toml, go.mod), чтобы определить язык стека. Затем сканирует папки ${CLAUDE_PLUGIN_ROOT} или .claude/ и собирает список установленных компонентов: субагентов, слэш-команд, хуков и навыков.
Обучение разбито на пять шагов. Ассистент объясняет роли субагентов вроде orchestrator, code-reviewer и debugger, показывает синтаксис хуков с кодами выхода 0 и 2, а также демонстрирует вызовы команд вроде /project-starter:commit. Для работы команде разрешены только инструменты Read, Glob и Grep, поэтому код и файлы проекта останутся в исходном виде.
Какие задачи решает
- разобраться в структуре установленных субагентов и навыков
- узнать логику срабатывания встроенных хуков и перехватчиков
- подобрать команды под конкретный язык и фреймворк репозитория
- изучить возможности плагина без риска перезаписать проектные файлы
Примеры запросов агенту
- /tutorial
- Проведи интерактивный туториал по плагину через /tutorial
- Покажи обзор агентов и хуков с помощью /tutorial
Как установить tutorial
Вручную: положить файлы
.claude/commands/tutorial.mdВызов — /tutorial. Для всех проектов — ~/.claude/commands/
Вручную: положить файлы
~/.codex/prompts/tutorial.mdВызов — /prompts:tutorial. В Codex этот формат помечен устаревшим, замена — скиллы
Вручную: положить файлы
.github/prompts/tutorial.prompt.mdВручную: положить файлы
.gemini/commands/tutorial.tomlGemini хранит команды в TOML: текст запроса — в поле prompt
Вопросы о tutorial
Как установить tutorial в Claude Code?
Поместите файл команды по пути .claude/commands/tutorial.md. После этого вызов /tutorial станет доступен в терминале Claude Code.
Изменяет ли команда tutorial файлы в репозитории?
Нет, это режим ознакомления только для чтения. Команда использует исключительно инструменты Read, Glob и Grep, поэтому файлы не модифицируются.
Как tutorial подстраивается под проект?
На шаге инициализации команда проверяет наличие файлов вроде package.json, pyproject.toml или go.mod. На основе найденного стека ассистент выдаёт индивидуальные рекомендации по командам и навыкам.
Исходный текст
tutorial.md · MIT
---
allowed-tools: Read, Glob, Grep
description: Interactive tutorial for learning the plugin. Walks through agents, skills, hooks, and commands with hands-on examples.
---
# Plugin Tutorial
Welcome the user and guide them through the plugin's features interactively. This is a read-only walkthrough -- no files are modified.
## Step 0: Detect Context
Before starting, silently gather context:
1. Use Glob to find `package.json`, `pyproject.toml`, `Cargo.toml`, `go.mod`, or other project markers in the working directory
2. Identify the project's primary language and framework
3. Use Glob to list the plugin's available components. They live in the plugin's install directory, `${CLAUDE_PLUGIN_ROOT}` (if that placeholder still appears literally here, the workflow was installed with npx: use `.claude/` instead):
- `${CLAUDE_PLUGIN_ROOT}/agents/*.md` for agents
- `${CLAUDE_PLUGIN_ROOT}/skills/*/SKILL.md` for skills
- `${CLAUDE_PLUGIN_ROOT}/hooks/hooks.json` for hooks
- `${CLAUDE_PLUGIN_ROOT}/commands/*.md` for commands
Then begin the tutorial.
## Step 1: Meet the Agents
Greet the user and introduce the agent system.
Read each file in `${CLAUDE_PLUGIN_ROOT}/agents/*.md` and present a summary table:
```
Welcome to the Claude Workflow Plugin tutorial.
Let's explore what this plugin gives you. First up: Agents.
Agents are specialized assistants that activate automatically based on
your request. Here's what's available:
| Agent | Triggers When You... | Model |
|-----------------|----------------------------------------|--------|
| orchestrator | Need multi-file coordinated changes | opus |
| code-reviewer | Ask for a code review | sonnet |
| debugger | Report a bug or error | sonnet |
| ... | ... | ... |
Agents marked PROACTIVELY in their description will activate
automatically when your prompt matches their trigger keywords.
You don't need to invoke them explicitly.
```
Explain that the orchestrator can spawn subagents for parallel work, and how agent selection works based on prompt keywords.
**Try it**: Suggest the user try a prompt like "review the last commit for issues" to see the code-reviewer agent activate.
## Step 2: Commands at Your Fingertips
Read the list of command files from `${CLAUDE_PLUGIN_ROOT}/commands/*.md` and present them:
```
Commands are slash-invocable workflows. Think of them as recipes
that combine multiple steps into one action.
Available commands:
| Command | What It Does |
|---------------------|-------------------------------------------|
| /project-starter:commit | Auto-generate conventional commit |
| /project-starter:verify-changes | Multi-agent verification suite |
| /project-starter:review | Code review with structured output |
| ... | ... |
Commands can be simple (commit) or complex (verify-changes spawns
5+ parallel subagents).
```
Read `${CLAUDE_PLUGIN_ROOT}/commands/commit.md` and walk through how it works:
```
Let's look at how /project-starter:commit works:
1. It gathers context: current branch, staged changes, recent commits
2. It analyzes the diff to understand what changed
3. It generates a conventional commit message (feat/fix/docs/...)
4. It executes the commit
Notice the frontmatter:
allowed-tools: Bash(git status:*), Bash(git diff:*), ...
This pre-approves those git commands for the turn, so the
command runs without permission prompts. It does not
restrict Claude to them.
```
**Try it**: Suggest staging a file and running `/project-starter:commit`.
## Step 3: Hooks That Protect Your Code
Read `${CLAUDE_PLUGIN_ROOT}/hooks/hooks.json` and explain each hook:
```
Hooks run automatically before or after certain actions.
They protect your code without you thinking about it.
Active hooks:
| Hook | Triggers On | What It Does |
|--------------------|-------------------|---------------------------|
| format-on-edit | File save/edit | Auto-formats changed files|
| ... | ... | ... |
Hooks use exit codes to control behavior:
Exit 0 = allow the action to proceed
Exit 2 = block the action and show a message
This means a hook can prevent you from committing secrets,
saving malformed config, or other mistakes.
```
Read one hook script to show the pattern:
```
Here's a simplified hook structure:
1. Receive input as JSON from stdin
2. Inspect the content or action
3. Exit 0 (allow) or exit 2 (block with message)
Hooks fail silently on errors (exit 0) so they never block
your workflow unexpectedly.
```
**Try it**: Suggest the user edit a file and observe the format-on-edit hook in action (if applicable to their project).
## Step 4: Skills That Enhance Responses
Read `${CLAUDE_PLUGIN_ROOT}/skills/*/SKILL.md` files and present the skill domains:
```
Skills inject domain knowledge into responses. When a skill's
trigger keywords match your prompt, its instructions are loaded
automatically.
Available skill domains:
| Skill Domain | Enhances Responses About... |
|-----------------|----------------------------------------|
| [skill-name] | [description from SKILL.md] |
| ... | ... |
Skills are passive -- they don't run commands. They shape how
the assistant thinks about your problem by providing specialized
knowledge, patterns, and best practices.
```
**Try it**: Suggest a prompt that would activate one of the skills relevant to their detected project type.
## Step 5: Your Personalized Next Steps
Based on the project context detected in Step 0, provide tailored recommendations:
```
Based on your project ([language/framework]), here's what I'd
recommend trying first:
1. **[Most relevant command]** -- [why it's useful for their stack]
Run: /project-starter:[command-name]
2. **[Most relevant agent]** -- [scenario where it helps]
Just describe your task and it will activate automatically.
3. **[Most relevant skill]** -- [how it will help]
Ask about [topic] and see enhanced responses.
Quick reference card:
/project-starter:commit -- commit with auto-message
/project-starter:verify-changes -- verify before pushing
/project-starter:review -- get a code review
/project-starter:bootstrap-repo -- generate full codebase docs
All available commands: /project-starter:[tab] to see the list.
```
## Presentation Guidelines
- **Conversational tone**: Write as if explaining to a colleague, not writing documentation.
- **Concise**: Each step should take 30 seconds to read. No walls of text.
- **Concrete examples**: Every step ends with a "Try it" suggestion.
- **Adaptive**: Tailor language and examples to the detected project type.
- **Progressive**: Each step builds on the previous one. Don't reference concepts before introducing them.
- **No modifications**: This command is read-only. Do not create, edit, or delete any files.
© CloudAI-X · MIT · CloudAI-X/claude-workflow-v2