本文へ移動
cccskills
無料GitHub で公開

explain

Explain code, concepts, or system behavior with adjustable depth levels

インストール方法を見る

含まれるファイル(1)

  • SKILL.md3.8 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Code explainer

Explain code, concepts, or system behavior with adjustable depth levels.

Purpose

Get clear explanations of:

  • How specific code works
  • Why certain patterns are used
  • What a system/module does
  • Architectural decisions and trade-offs

Instructions

Step 1: Determine scope

Identify what needs explanation:

  • File: Entire file structure and purpose
  • Function/Method: Specific implementation details
  • Concept: Architectural pattern or design decision
  • Flow: How data/control moves through the system

Step 2: Assess complexity

Simple (1-2 min read)     → Quick summary, key points only
Standard (3-5 min read)   → Purpose, how it works, key decisions
Deep (10+ min read)       → Full breakdown, alternatives, trade-offs

Step 3: Gather context

# For file explanations
head -50 "$FILE"  # See imports and structure

# For function explanations
grep -A 30 "function $NAME\|def $NAME\|fn $NAME" "$FILE"

# For module explanations
ls -la "$DIR"
cat "$DIR/index.ts" 2>/dev/null || cat "$DIR/__init__.py" 2>/dev/null

Step 4: Structure the explanation

Output format


📖 Explanation: [Target]

Scope: [file/function/concept/flow] Depth: [simple/standard/deep]

What it does

[1-3 sentences describing the purpose]

How it works

[Step-by-step breakdown appropriate to depth level]

Key decisions

DecisionWhyAlternative
[choice made][reasoning][what else could work]

Example usage

// How to use this correctly

Related code

  • path/to/related.ts - [relationship]
  • path/to/dependency.ts - [relationship]

💡 Learning Notes (if --learn flag)

[Additional context for understanding the broader pattern]


Depth levels

Simple (/explain --simple)

**validateUser()** checks if the user object has required fields
(email, password) and returns a boolean. Uses regex for email format.

Standard (/explain - default)

**validateUser(user: User): ValidationResult**

**Purpose**: Validates user input before database operations.

**Flow**:
1. Check required fields exist (email, password)
2. Validate email format with regex
3. Check password meets requirements (8+ chars, special char)
4. Return { valid: boolean, errors: string[] }

**Used by**: signup(), updateProfile()

Deep (/explain --deep)

[All of standard, plus:]

**Design Decisions**:
- Returns ValidationResult instead of throwing to allow batch validation
- Regex chosen over library for zero dependencies
- Password rules configurable via config.ts

**Trade-offs**:
- Pro: Fast, no dependencies
- Con: Regex email validation isn't RFC-compliant

**Alternatives Considered**:
- Zod schema: More powerful but adds 50KB
- Class-validator: Better for decorators but OOP-heavy

Usage examples

Explain a file:

/explain src/auth/middleware.ts

Explain a function:

/explain the handleWebhook function in payments.ts

Explain a concept:

/explain how our event sourcing works

Explain with specific depth:

/explain --deep the authentication flow
/explain --simple what useCallback does

Explain for learning:

/explain --learn the repository pattern used here

Tips

  1. Be specific: "Explain line 45-60" > "Explain this file"
  2. State your level: "I'm new to TypeScript" helps calibrate
  3. Ask follow-ups: "Why not use X instead?" deepens understanding
  4. Request analogies: "Explain like I'm familiar with Python but not TS"

$ARGUMENTS

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Audit Claude Code agents, skills, and commands for quality and production readiness. Use when evaluating skill quality, checking production readiness scores, or comparing agents against best-practice templates.

日本語の概要は準備中です。原文の説明を表示しています。

FlorianBruniaux/claude-code-ultimate-guide6,1432026年10月7日 更新

Codebase health audit scoring 7 categories with progression plan

日本語の概要は準備中です。原文の説明を表示しています。

FlorianBruniaux/claude-code-ultimate-guide6,1432026年10月7日 更新

Autonomous improvement loop: scan codebase metrics, scaffold experiment files, run agent-driven iterations until metric improves

日本語の概要は準備中です。原文の説明を表示しています。

FlorianBruniaux/claude-code-ultimate-guide6,1432026年10月7日 更新

best-of-n

無料

Generate bounded independent candidates, score them against a frozen rubric, and verify the selected result with a proof log.

日本語の概要は準備中です。原文の説明を表示しています。

FlorianBruniaux/claude-code-ultimate-guide6,1432026年10月7日 更新

canary

無料

Post-deploy monitoring: watch production after a deploy and alert on regressions

日本語の概要は準備中です。原文の説明を表示しています。

FlorianBruniaux/claude-code-ultimate-guide6,1432026年10月7日 更新

catchup

無料

Restore context after /clear by summarizing recent work and project state

日本語の概要は準備中です。原文の説明を表示しています。

FlorianBruniaux/claude-code-ultimate-guide6,1432026年10月7日 更新

FlorianBruniaux のスキルをすべて見る

このスキルの問題を報告する