ワンクリックで
planning
Planning a change to any codebase. Use when user asks you to plan something instead of taking immediate action.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
Planning a change to any codebase. Use when user asks you to plan something instead of taking immediate action.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Review a GitHub pull request for correctness, security, observability, test coverage, and conventions. Identify applicable skills, verify and triage sub-agent findings to cut noise, score each review dimension purple-red-amber-green, and produce a structured review saved to disk and presented to the user. Use when user asks to review a PR, check a pull request, or give feedback on changes in a PR.
Walk a user through a GitHub pull request to help them understand what it does, using markdown visualisations and a guided diff-by-diff tour. Use when the user wants to understand, explore, learn from, or be walked through a PR (not when they want a formal review).
Capture aha-moments as durable insight notes under `.ai/insight/`, or quiz the user on past insights to reinforce recall. Use when the user has just learnt something non-obvious (a fix they didn't know, an explanation that clicked, a gotcha discovered) and wants to record it, or when the user invokes `/aha quiz` to be tested on stored insights.
Toggle persistent planning mode on/off. When on, hook support can remind the agent to keep planning documents updated throughout the conversation, with manual fallback if unavailable. Combines the full planning workflow with persistent mode for long-running planning and implementation sessions. Use with a topic to start, with "continue" to resume an existing plan, or with "off" to stop.
Toggle persistent research mode on/off. When on, hook support can remind the agent to keep the research document updated throughout the conversation, with manual fallback if unavailable. Use with a research topic to start, with "continue" to resume an existing research doc, or with "off" to stop.
Retrieve saved plan folders from `.ai/planning`, let the user choose one, then load its planning documents and summarise implementation status. Use when user asks to fetch, reopen, resume, continue, or inspect an existing plan.
| name | planning |
| description | Planning a change to any codebase. Use when user asks you to plan something instead of taking immediate action. |
| argument-hint | ["the work to plan"] |
{repo root}/.ai/planning/{yyyy-mm-dd} {task-name}/ - where:
{repo root} is the root of the current repository{yyyy-mm-dd} is the date of the plan creation{task-name} is an extremely succinct name for the task in kebab-caseUse persistent markdown files as your "working memory on disk." Context windows are volatile and limited; anything important gets written to disk.
./templates/| File | Purpose | When to Update | Scope |
|---|---|---|---|
plan.md | Work breakdown, decisions | After each phase | All plans |
findings.md | Research, discoveries | Per 2-action rule | All plans |
questions.md | Log of questions and user responses | Throughout planning | All plans |
progress.md | Session log, test results | Throughout session | Multi-phase only |
If you decide or are expected to produce a plan, follow the steps outlined in this section in order.
{task-name} — an extremely succinct name for the task, use kebab-case e.g. refactor-mfa-flowfindings.md following the template: templates/findings.mdfindings.md as you goBased on your research, explicitly determine whether this is a multi-phase plan. A plan is multi-phase if it meets ANY of the following:
Document your determination and reasoning at the top of findings.md under a ## Plan Size heading.
Create the remaining planning files from the templates based on the determination in step 2 (see Planning files). Populate plan.md with the plan and questions.md with any unresolved questions. Keep questions.md scaffolded even if you have no questions yet, as they may come up later. For multi-phase plans, also create progress.md as a stub — it will be populated during execution.
/rename {task-name} to rename this conversation."/name {task-name} to name this conversation."{task-name} if your tool supports it."planning skill (in each phase, for multi-phase plans)rust-bevy-standards, java-conventions, tdd)tdd skill for any code you write and task descriptions featuring development work must make this clear"After every 2 view/browser/search operations, IMMEDIATELY save key findings to text files."
This prevents visual/multimodal information from being lost.
Before major decisions, read the plan file. This keeps goals in your attention window.
After completing any phase:
## Status in plan.mdIn progress → CompleteEvery error goes in plan.md (the single home for errors across all planning files). This builds knowledge and prevents repetition.
if action_failed:
next_action != same_action
Track what you tried. Mutate the approach.
ATTEMPT 1: Diagnose & Fix
→ Read error carefully
→ Identify root cause
→ Apply targeted fix
ATTEMPT 2: Alternative Approach
→ Same error? Try different method
→ Different tool? Different library?
→ NEVER repeat exact same failing action
ATTEMPT 3: Broader Rethink
→ Question assumptions
→ Search for solutions
→ Consider updating the plan
AFTER 3 FAILURES: Escalate to User
→ Explain what you tried
→ Share the specific error
→ Ask for guidance