AGENT SKILL

plan

Write the plan file in docs/plans/ that code changes get before building, unless the user has said to build without one. Use when the user asks to build, fix, edit or change code and docs/plans/ has no approved plan for it, or when a plan must be revised after review. Plan step of the methodical loop. Not for questions or reading code, and not when the user has waived the plan for this change (go to work).

License 许可
MIT
最近更新
2026-11-15
Stars
★ 0
Forks
0

01 前言 Front Matter

name
plan
description
Write the plan file in docs/plans/ that code changes get before building, unless the user has said to build without one. Use when the user asks to build, fix, edit or change code and docs/plans/ has no approved plan for it, or when a plan must be revised after review. Plan step of the methodical loop. Not for questions or reading code, and not when the user has waived the plan for this change (go to work).

03 SKILL.md 正文

Plan

Say Step: plan before anything else.

What it produces

docs/plans/<YYYY-MM-DD>-<slug>.md, made from ${PLUGIN_ROOT}/templates/plan.md, with every section filled or explicitly marked "none".

The plan's base: is the commit it was written against. Why: a reviewer can only check a plan against the code it was written for, and a stale base is itself a reason to deepen.

The plan is a deciding seat, written on the decider's row, which stays the decider's row at every escalation level. If this session runs on another model, launch a general-purpose Agent with the decider's model and effort to draft it, and give that seat the base commit and the brainstorm. If that launch fails, follow the re-seat ladder in ${PLUGIN_ROOT}/reference/seat-selection.md.

How

  1. Take the inventory from the code, not from memory. Every file, module, interface and document the unit touches, with what changes.

  2. Look for prior art in docs/solutions/ and name what applies.

  3. Score deepen against the always-on rules' triggers, fill the Deepen score section, and say the score aloud ("Deepen score: 3 of 6, deepening; escalation: lean"). Below 2, run it anyway if it would help, and say why. Why: saying the score lets the user overrule it.

    Set the escalation level. This step computes the line once and writes it as escalation: in the front matter: full at a score of 4 or more, lean at 3 or at a named trigger (Escalation in ${PLUGIN_ROOT}/reference/seat-selection.md), otherwise none. The Deepen score section records which trigger set the level, with its evidence. If the user edits the line before approving, the user's edit stands and is recorded as a ruling; a deepen or review revision recomputes only when no such edit is recorded. Work reads the approved plan's line. Why: the level is decided where the risk is known, and the user's edit outranks the computation.

  4. Write acceptance tests the person served would recognise, each with how it is run and what counts as a pass.

  5. Record rulings in the Rulings section as they are given, after you read ${PLUGIN_ROOT}/rules/user-words.md and follow it.

  6. Commit the plan. If the repo is on main, first create the unit's branch (see the work skill), so the plan and its work merge together.

When it hands on

  • Score 2 or more, or a stale base: say Step: deepen and continue with the deepen skill in the same turn.
  • Otherwise: say Step: plan-review and run /methodical:plan-review in the same turn.

Leave the plan at state: draft. Plan review sets reviewed. The question "Approve this plan?" comes next: plan-review asks it with the native structured-question tool, as its own final call after the open-question answers are in the plan; record their answer in Rulings, and set state: approved only on their yes. Why: the user, not a reviewer, rules that a plan gets built; that stop is what makes the method gated. Open questions wait for their ruling while everything that does not depend on them continues.

04 同仓库更多 Skills

BigLeverVentures
BigLeverVentures/methodical-gated-engineering

phrase-outside

暂无简介

★ 0
BigLeverVentures
BigLeverVentures/methodical-gated-engineering

named-only

暂无简介

★ 0
BigLeverVentures
BigLeverVentures/methodical-gated-engineering

work

Build an approved plan from docs/plans/ on its own short-lived branch, or a change the user said to make without a plan. Use when a plan reads state: approved and the user wants it built, when resuming such a build, or when the user has waived the plan for a named change. Work step of the methodical loop. Not when no approved plan exists and the user has not waived one (run the plan step), and not for reviewing finished work.

★ 0
BigLeverVentures
BigLeverVentures/methodical-gated-engineering

work-review

Review a finished build on its branch against its plan, with reviewer seats matched to the code's language, before it merges. Use when the work step's tests pass and the plan's approach is done, or when the user asks for a review of a branch's changes. Work-review step of the methodical loop. Not for reviewing a plan.

★ 0
BigLeverVentures
BigLeverVentures/methodical-gated-engineering

plan

Write the plan file in docs/plans/ that code changes get before building, unless the user has said to build without one. Use when the user asks to build, fix, edit or change code and docs/plans/ has no approved plan for it, or when a plan must be revised after review. Plan step of the methodical loop. Not for questions or reading code, and not when the user has waived the plan for this change (go to work).

★ 0
BigLeverVentures
BigLeverVentures/methodical-gated-engineering

plan-review

Review a plan in docs/plans/ with reviewer seats matched to the project's language, before the user approves it. Use when a plan has just been written or deepened, when a plan was revised after review, or when the user asks for a plan review. Plan-review step of the methodical loop. Not for reviewing built code.

★ 0