# Principle Plan Present Wait

> Requires a written plan and user approval before mutations. Apply when consequential changes need authorization.

- Skill: `bostonaholic/principle-plan-present-wait` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add bostonaholic/principle-plan-present-wait`
- Raw SKILL.md: https://api.skillmd.com/api/skills/bostonaholic/principle-plan-present-wait/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: bostonaholic (https://skillmd.com/u/bostonaholic)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/bostonaholic/principle-plan-present-wait

---


# Plan, Present, Wait

Write mutation plans before presenting one question and one recommendation per consequential choice; Nothing changes before the user answers, and no answer means no mutation outside the verified-confidence carve-out.

- Keep ask and act in separate turns; execute only answered items from partial answers.
- Persist plans that may survive a turn or compaction, then re-read them before execution; use conversation lists only for same-session punch lists.
- Present irreversible mutations as exact output text, one consequential choice per question.
- Approve reversible classes only with stated undo; name every target and its evidence.
- Re-validate each step against the approved class; approval never relaxes hard rules.
- Skip waiting only above a verified confidence bar and within every hard rule; always present below-bar or carve-out items.