Skills catalog

ABC commands open the doors. Skills are the steps underneath — invoke one at a time when you want to redo a step or watch what it does.

Context

Skill	What it does
/explore	Agent setup, system architecture, conceptual model, and PRD shell — from the repo tree and Guide files only.
/extract	Per-container architecture, schemas, and coding rules — one container at a time, reading source.

Skill	What it does
/specify	A feature spec or a refactor spec — kind named by the caller. Stops for you to read it.

Skill	What it does
/planify	One implementation plan per container from the approved spec.
/codify	The only skill that writes code — source, unit tests, and the e2e suite.

Skill	What it does
/verify	E2e verdict against acceptance criteria — report only, never fixes.
/qualify	Quality gates — report only. Failed gates loop back through /codify.

Skill	What it does
/release	Version, changelog, reconciled docs, and tag — only after verified.

Architect, Builder, Craftsman — each command chains skills in fresh subagents so no step inherits the previous step's clutter.

Command	What it does
`/architect-map`	Architect: map what is there — architecture, schemas, coding rules. Works on empty and mature repos.
`/builder-ship`	Builder: turn a requirement into a spec, then plan, build, verify, and ship.
`/craftsman-refactor`	Craftsman: re-map for drift or take your proposal, then ship the fix through /ship-spec.

You only review twice. Everything else runs without asking.|

Trigger	You check
After /architect-map	Architecture, schemas, and coding rules match the repo
After /builder-ship spec	Problem, outcomes, and acceptance criteria are right
After /craftsman-refactor	The defect or proposal is right before /ship-spec runs

flowchart LR
  A[architect-map] -->|you ✔️| B[builder-ship]
  B -->|you ✔️| S[ship]
  S -->|you ✔️| C[craftsman-refactor]