/we:vision — Vision (Solo) at the PRD altitude

You sharpen or create one PRD at docs/plans/<vision>/PRD.md: who the product is for, what is broken
for them, what changes when it works, what we will not build. One PRD per product, always Markdown, never a ticket.
Vision Solo never decomposes: Sagas come from /we:meet vision, each Saga is formulated by /we:saga.
Wanted stops: one interview question at a time, and the approval. After the approval you commit, print the next verb and stop.

Shared contract (altitudes, drafting, writer, commit path): ${CLAUDE_PLUGIN_ROOT}/references/apo-hierarchy.md.
Template: ${CLAUDE_PLUGIN_ROOT}/skills/vision/references/template.md.

Frame — the four questions

Then the bets and assumptions the answers rest on, each falsifiable. A PRD that changes every quarter
is a Saga in disguise: say so. Feature ideas that surface belong to a Saga or Epic; note them for
/we:meet vision and return to the frame.

Steps

1. Load. The PRD and its folder (docs/plans/<vision>/), docs/plans/out-of-scope/ if present
   (already-decided non-bets: cite them, do not re-open them), the instruction files, and the existing
   Sagas (docs/plans/*-saga.md with vision: <vision>). With a materialised Companion, read its goals
   (list_goals) and name a tension between them and the PRD; never resolve it silently.

2. Interview per apo-hierarchy.md § Drafting step 1, walking the four questions. On Refine, read
   back each current answer in your words and ask what has shifted; hedge words ("mostly", "kind of")
   mark the fuzzy edge.

3. Draft with a we:dev-high writer per § Drafting step 3. The writer rewrites the whole PRD; it is
   short enough.

4. Approval per § Drafting step 4, showing the PRD in full.

5. After approval: set updated: to today, commit per § Drafting step 2, then output and stop:

   PRD: docs/plans/<vision>/PRD.md (<sha> on <branch>).
   Next: /we:meet vision — validate the PRD and derive the Sagas.