to-plan

name	to-plan
description	After grill-me (or equivalent resolved decisions), produce a concise execution plan for the next agent session. No interviews, no issue tracker, no requirement expansion.

Package locked decisions from the conversation into a short execution plan. Do not plan for the user — only record what was already decided.

Prerequisite

Grill-me (or equivalent) is finished: every branch resolved. If open questions remain, stop and tell the user to resume grilling.

Do not

• Interview the user or ask for approval on scope
• Invent requirements, user stories, or alternatives
• Sketch modules for sign-off or propose architecture not already agreed
• Publish to an issue tracker or apply triage labels

Do

1. Read the conversation for locked decisions only
2. Minimal codebase lookup — name real paths/packages to touch (not exploratory architecture)
3. Emit one markdown artifact using the template below (in chat, or a file if the user asks)

Goal

One or two sentences.

Decisions

Bullet list from grill-me — locked, no debate.

Out of scope

What was explicitly excluded.

Touch

Paths and packages (concrete).

Steps

Ordered checkboxes — implementation sequence only.

Skills per step

Which skills to read fully per step (e.g. component-scaffold, tdd, browser-tests).

Verify

Checkers from AGENTS.md#code-quality-checkers (and Other packages when applicable).

When done, the user or next agent can start implementation (typically tdd + design-system skills).