# Plan

> Turn a sufficiently understood task into an ordered implementation plan with dependencies and verification. Orchestrate Codewhale’s native plan state; do not build a parallel planner.

- Skill: `hmbown/plan` (Agent Skill)
- Install (CLI): `npx skillmds@latest add hmbown/plan`
- Raw SKILL.md: https://api.skillmd.com/api/skills/hmbown/plan/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Productivity
- Author: Hmbown (https://skillmd.com/u/hmbown)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/hmbown/plan

---


# Plan

## When to use
Use when the task is understood well enough to sequence work, but not yet a single obvious next edit.

## Non-goals
- Do not plan while key goals are still unknown — use interview first.
- Do not create a second planner outside Codewhale plan/Work state.

## Workflow
1. Restate the goal and constraints.
2. List ordered steps with dependencies.
3. Mark verification for each risky step.
4. Write the plan into Codewhale’s native plan/Work surface.
5. Wait for authorization before broad implementation.

## Output shape
- Goal
- Steps (ordered)
- Verification
- Risks / open questions