Create an implementation task plan at _/local-plans/<plan-name>.md .
Argument handling
- With arguments → Use the arguments as the task description and start creating the plan
- Without arguments → Ask the user for the task details before starting
Principles of action
- Investigate the codebase first — Do not make plans based on guesswork. Use Explore Agent and Glob/Grep to read existing code and understand the current state
- Write at a verifiable granularity — The completion conditions for each step must be clear
- Always include the file structure — Clearly specify the files to be created and modified in tree format
- Keep phases independent — Divide into units that can be executed in parallel as much as possible
- Write in Japanese — Create the plan file in Japanese
- Writing to locations that require authentication should be done via
_/performed via —.claude/When creating files in directories that trigger authentication prompts, first create them temporarily in_/, and after completion, move them to the target location withmv. During cleanup, delete only the files and subdirectories you created, and only if_/is empty, delete it withrmdir(rm -rf _/dotclaudeis prohibited — it may delete files from parallel work)
Plan creation procedure
Step 1: Understanding the task
- Analyze the user's explanation
- Investigate the codebase as needed (Explore Agent / Glob / Grep)
- Structure of related existing files
- Technology stack and libraries in use
- Existing similar implementations and patterns
- Scope of impact (files affected by changes)
- Organize preconditions and constraints
Step 2: Drafting the plan
Structure the plan according to the standard format below.
Step 3: Creating the plan file
- File name: Use a concise name in kebab-case that reflects the task content
- Example:
auth-middleware-rewrite.md,chakra-ui-skill.md,api-endpoint-refactor.md
- Example:
- Output destination:
_/local-plans/<plan-name>.md - Report the plan file path to the user
Standard format (template)
The plan must be in a format verifiable by plan-verifier Agent.
Write the file structure with full paths and clearly specify the deliverables for each phase.
# {タスクタイトル}
## 背景・目的
{なぜこのタスクが必要か}
{解決したい課題}
{期待される成果}
## 現状整理
{関連する現在の状態 — 既存ファイル、技術スタック、依存関係}
## 設計
### ファイル構成
{作成・変更するファイルのツリー図}
### 仕様
{技術的な設計詳細 — API、データ構造、フォーマット等}
## 実装ステップ
### Phase 1: {フェーズ名}
1. {具体的なタスク}
2. {具体的なタスク}
### Phase 2: {フェーズ名}
1. {具体的なタスク}
2. {具体的なタスク}
{必要に応じて Phase N まで}
## 並列実行戦略
{並列化可能な作業がある場合のみ記載}
## 検証方法
{計画の成果物が正しく完成したことをどう確認するか}
{plan-verifier Agent での検証ポイント}
## 将来の拡張
{スコープ外だが関連する発展可能性 — 任意}
Notes for each section
- Background and purpose: Required. Clarify why it is being created
- Current state: Describe any related existing state. Can be omitted if only creating something new
- Design: Required. Always include the file structure (in tree format)
- Implementation Steps: Required. Divide into phases, and make each step a concrete action
- Parallel Execution Strategy: Include only if there is work that can be parallelized
- Verification Method: Required. Clearly state how to verify the deliverables
- Future Extensions: Optional. Include any related matters outside the scope