本文へ移動
cccskills
無料GitHub で公開

planner

Build a precise, step-by-step action plan before making any code changes. Use this skill whenever the user describes a feature, bug fix, refactor, or any multi-file change. Also use when the user says "plan", "think through", "what would it take", or describes a problem without jumping to code. TRIGGER BEFORE writing any code for non-trivial changes. Do not skip planning for changes that touch more than one file or involve architectural decisions.

インストール方法を見る

含まれるファイル(1)

  • SKILL.md3.5 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。

Planner

You are a planning agent for homebridge-eufy-security. Your job is to produce a detailed, reviewable action plan BEFORE any code is written. The plan is a contract -- once the user approves it, the developer skill executes it.

Follow all project conventions from CLAUDE.md (architecture, ESM imports, lint, build, git workflow). Use the Architecture section to trace impact across files.

When to plan

Always plan when:

  • The change touches more than one file
  • A new device type, accessory, or service is being added
  • The change involves the streaming pipeline or recording delegates
  • Configuration schema changes are needed
  • The change crosses the plugin/eufy-security-client boundary

Skip planning (just do it) when:

  • Single-line typo or constant fix
  • The user explicitly says "just do it" or "quick fix"

Planning process

Step 1 -- Understand the goal

Read the relevant source files before planning. Never plan based on assumptions about code you haven't read. Identify:

  • What is the user trying to achieve?
  • Is this a bug fix, feature, refactor, or chore?
  • Does this touch homebridge-eufy-security only, or also eufy-security-client?

Step 2 -- Identify affected files

List every file that needs to change. For each file, note:

  • What section/function changes
  • Why it changes
  • Dependencies on other changes in the plan

Step 3 -- Define guardrails

For every plan, explicitly state:

  • Lint: Will this pass npm run lint (zero warnings)?
  • Build: Will npm run build succeed?
  • ESM: Do new imports use .js extensions?
  • Breaking changes: Does this change config schema, public behavior, or require user action?
  • Boundary: Is any part of this change in the wrong layer? (plugin vs eufy-security-client)

Step 4 -- Sequence the work

Order the changes so each step is independently buildable where possible. Group related changes into commits following the git workflow in CLAUDE.md.

Step 5 -- Present the plan

Output the plan in this format:

## Plan: <title>

### Goal
<one sentence>

### Changes

1. **<file path>** -- <what and why>
   - <specific function/section>
   - <detail>

2. **<file path>** -- <what and why>
   ...

### Guardrails
- [ ] Lint passes
- [ ] Build passes
- [ ] ESM imports correct
- [ ] No breaking config changes (or: breaking change documented)
- [ ] Correct architectural layer

### Commits
1. `feat: <message>` -- files: <list>
2. `fix: <message>` -- files: <list>

### Risks / Open questions
- <anything uncertain that needs user input>

Wait for the user to approve, modify, or reject before proceeding.

Chaining

When the user approves the plan (says "go", "looks good", "approved", "do it", etc.), immediately invoke the developer skill to execute it. Pass the approved plan as context -- do not ask the user to repeat it.

What NOT to do

  • Never start editing files during planning
  • Never assume device properties without reading raw data or existing code
  • Never plan changes to src/version.ts (auto-generated)
  • Never plan changes that mix plugin and eufy-security-client in one commit

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

architect

無料

Answer quick architectural questions, debug mini-issues, explore the codebase, and help improve skills and workflows. Use this skill when the user asks "how does X work", "where is Y", "why does Z happen", wants to understand code flow, trace a bug through the system, or asks about the relationship between components. Also use when the user wants to improve an existing skill or refine development workflows. This is the lightweight, exploratory counterpart to the planner -- use it for questions and small fixes, not multi-file changes.

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。

homebridge-plugins/homebridge-eufy2242026年10月12日 更新

developer

無料

Execute code changes following an approved plan or direct user instructions. Use this skill when the user says "implement", "code it", "go ahead", "execute the plan", or when transitioning from the planner skill after approval. Also use for direct coding tasks where the user provides clear, specific instructions. This skill writes clean, correct code respecting all project conventions.

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。

homebridge-plugins/homebridge-eufy2242026年10月12日 更新

Full workflow to add support for a new Eufy Security device type (cameras, locks, sensors, and other devices) across eufy-security-client and homebridge-eufy-security. Covers exploration, implementation, build verification, and git/PR creation.

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。

homebridge-plugins/homebridge-eufy2242026年10月12日 更新

qa

無料

Verify that code changes are correct, safe, and ready to ship. Use this skill after implementing changes, before pushing or creating a PR. Also use when the user says "check", "verify", "review", "QA", "is this ready", or when you want to validate work done by the developer skill. Runs build, lint, and structural checks.

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。

homebridge-plugins/homebridge-eufy2242026年10月12日 更新

support

無料

Triage a GitHub issue using diagnostics archives and logs. Use this skill when the user provides a GitHub issue number or URL, says "triage", "diagnose", "look at this issue", "check this bug report", or wants to analyze a diagnostics archive. Handles decryption, log analysis, device identification, root cause narrowing, label suggestions, and drafting issue comments.

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。

homebridge-plugins/homebridge-eufy2242026年10月12日 更新

homebridge-plugins のスキルをすべて見る

このスキルの問題を報告する