# Planning

> Planning lifecycle: specs, requirements, ambiguity triage, human-source elicitation, evidence prototypes, file-backed plans, validation, pause/resume, and context boundaries.

- Skill: `notque/planning` (Agent Skill, multi-file: 22 files)
- Install (CLI): `npx skillmds@latest add notque/planning`
- Raw SKILL.md: https://api.skillmd.com/api/skills/notque/planning/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: notque (https://skillmd.com/u/notque)
- Updated: 2026-09-10
- Page: https://skillmd.com/skills/notque/planning

---


# Planning Skill

Select the planning method below. `/do` owns capability routing; these references are internal methods. Planning owns specs, saved plans, and plan pause/resume. `subagent-driven-development` owns execution; `session-handoff` owns inline transfers, including live processes. Use `verification-before-completion` for outcome checks.

## Reference Loading Table

| Signal | Load These Files | Why |
|---|---|---|
| "write spec", "user stories", "define requirements", "scope this", "acceptance criteria", "define scope", "spec out" | `${CLAUDE_SKILL_DIR}/references/spec.md` | **Spec** |
| "discuss ambiguities", "resolve gray areas", "clarify before planning", "assumptions mode", "before we plan", "pre-planning discussion" | `${CLAUDE_SKILL_DIR}/references/pre-plan.md` | **Pre-plan** |
| "interview me", "grill me", "depth-first review", "depth-first interview", "not sure", "i'm not sure", "where do i start", "want clarity on", "need clarity on", "what am i missing", "poke holes in", "challenge my assumptions", "think this through with me", "lots of moving parts", "many decisions" | `${CLAUDE_SKILL_DIR}/references/depth-first-interview.md` | **Interview** |
| implicit ambiguity or unclear implementation choices | `${CLAUDE_SKILL_DIR}/references/ambiguity-triage.md` | Decide whether questions earn their interruption cost |
| another person holds needed facts, preferences, constraints, or approval | `${CLAUDE_SKILL_DIR}/references/human-source-elicitation.md` | Create a targeted, sendable question artifact |
| observation can settle a disputed or unknown choice | `${CLAUDE_SKILL_DIR}/references/empirical-prototype.md` | Turn uncertainty into bounded evidence |
| a phase, worker, or session transition is near | `${CLAUDE_SKILL_DIR}/references/context-boundary.md` | Preserve only the state the receiver needs |
| "create plan", "task plan", "working memory", "persistent plan", "file-backed planning" | `${CLAUDE_SKILL_DIR}/references/plan-files.md` (+ `${CLAUDE_SKILL_DIR}/references/executor-ready-plan-template.md` when plan targets SDD execution) | **Plan-files** |
| "check plan", "validate plan", "plan checker", "review plan", "is this plan ready", "pre-execution check" | `${CLAUDE_SKILL_DIR}/references/check.md` | **Check** |
| "list plans", "show plan", "complete plan", "plan status", "manage plans" | `${CLAUDE_SKILL_DIR}/references/manage.md` | **Manage** |
| "pause", "save progress", "handoff", "stopping for now", "end session", "session handoff", "wrap up session" | `${CLAUDE_SKILL_DIR}/references/pause.md` | **Pause** |
| "resume", "continue", "pick up where I left off", "what was I doing", "continue work", "where did I leave off", "what's next" | `${CLAUDE_SKILL_DIR}/references/resume.md` | **Resume** |

For interviews, use independent questions batched into frontier rounds, dependent questions asked sequentially, and a recommendation per question.

## Instructions

1. Identify the planning intent and inspect available evidence.
2. For implicit ambiguity, load `${CLAUDE_SKILL_DIR}/references/ambiguity-triage.md` before choosing an interview.
3. Load the matching reference. Compose references when the task crosses sources or context boundaries.
4. Follow the selected reference and return to execution as soon as its gate clears.