/align
AlignmentUser-invoked

Make fuzzy decisions explicit before /reqs, /design, or /dev, and land shared vocabulary in CONTEXT.md.

Before /reqs, /to-prd, /design, or /dev when scope, terms, or trade-offs are still fuzzy.

  • Skills bootstrapped; /setup done if you will publish later.
  • A fuzzy idea, plan, or feature name — exact scope can wait.
  1. Invoke /align and name the feature or plan to grill.
  2. Answer one question at a time — pick A/B/C/D when offered (recommended option marked).
  3. Let the agent update `CONTEXT.md` with shared vocabulary as you go.
  4. When decisions are clear, follow the `## Next Step` cue (/reqs, /to-prd, /design, or /dev).

Major forks are settled, domain words live in `CONTEXT.md`, and the next pipeline skill is named.

Upstream
Idea or rough plan
Downstream
/reqs or /to-prd → /design → /dev

Does not write a PRD — use /reqs for a full draft or /to-prd to publish a lean one. Does not implement code.

  • Still need discovery or a long PRD? Next is /reqs. Aligned and ready to publish lean? Use /to-prd.
  • The grill stays in this chat — there is no `align` subagent.

Uses align-loop and domain-modeling under the hood (model-invoked). The grill stays in this chat — no subagent.

Sample prompt

prompt
/align

Grill the plan for [feature].