Files
antigravity-claudekit/skills/ck-planning/SKILL.md
2026-02-16 14:02:42 +09:00

107 lines
3.5 KiB
Markdown

---
name: ck-planning
description: Plan feature implementations, design system architectures, create technical roadmaps. Use for implementation planning, system design, solution architecture, phase documentation, evaluating technical trade-offs.
---
# ck-planning
Create detailed technical implementation plans through research, codebase analysis, solution design, and comprehensive documentation. Embeds the `planner` agent role.
## Core Principles
Always honor **YAGNI**, **KISS**, and **DRY**. Be honest, brutal, straight to the point, and concise.
## When to Use
- Planning new feature implementations
- Architecting system designs
- Evaluating technical approaches
- Creating implementation roadmaps
- Breaking down complex requirements
- Assessing technical trade-offs
## Don't Use When
- A plan already exists and just needs execution — use `ck-cook` instead
- The task is trivial and doesn't require multi-phase planning
- You need research only, without creating a full plan — use `ck-research`
## Core Process
### 1. Research & Analysis
Dispatch parallel `ck-research` agents to investigate relevant technical topics. Skip if researcher reports are already provided.
### 2. Codebase Understanding
Search the codebase for relevant files, read docs in `./docs`. Skip if scout reports are already provided.
### 3. Solution Design
Evaluate multiple approaches, apply mental models:
- **Decomposition**: Break the epic into concrete stories
- **Working Backwards**: Start from desired outcome, identify every step
- **Second-Order Thinking**: Ask "And then what?" for hidden consequences
- **The 80/20 Rule**: Identify the 20% of features delivering 80% of value
- **Risk & Dependency Management**: What could go wrong? What does this depend on?
### 4. Plan Creation
Save plans to `./plans` directory using the naming pattern from the `## Naming` section injected by the context system. Pattern includes the full path and computed date.
**Plan directory structure:**
```
plans/
└── {date}-plan-name/
├── research/
├── reports/
├── plan.md
└── phase-XX-phase-name.md
```
**Every `plan.md` must start with YAML frontmatter:**
```yaml
---
title: "{Brief title}"
description: "{One sentence for card preview}"
status: pending
priority: P2
effort: {sum of phases}
branch: {current git branch}
tags: [relevant, tags]
created: {YYYY-MM-DD}
---
```
### 5. Phase Files
Each phase file must contain:
- Context Links
- Overview (priority, status, description)
- Key Insights
- Requirements (functional + non-functional)
- Architecture
- Related Code Files
- Implementation Steps (detailed, numbered)
- Todo List (checkbox)
- Success Criteria
- Risk Assessment
- Security Considerations
- Next Steps
## Active Plan State
Check the `## Plan Context` section in injected context:
- **"Plan: {path}"**: Active plan — ask "Continue with existing plan? [Y/n]"
- **"Suggested: {path}"**: Branch-matched hint — inform user, ask to activate or create new
- **"Plan: none"**: No active plan — create new using `## Naming` pattern
After creating plan folder, update session state so subsequent agents receive correct context.
## Output Requirements
- DO NOT implement code — only create plans
- Respond with plan file path and summary
- Ensure self-contained plans with all necessary context
- Include code snippets/pseudocode when clarifying architecture
- Provide multiple options with trade-offs when appropriate
- Make plans detailed enough for junior developers
**IMPORTANT:** Sacrifice grammar for concision. List unresolved questions at end if any.