spec-clarify

# spec-clarify — disciplined ambiguity reduction for PLAN-NNN > **Activation**: manual-only via `/spawn spec-clarify PLAN-<NNN>`. > Typical use: between `/debate round1` and `/debate round2` to reduce > ambiguity before adversarial round 2. > Never invoked by `/debate` or `/ceo-boot` by default. ## Purpose Port of github/spec-kit's `/speckit.clarify` (`templates/commands/clarify.md`). Implements a sequential 5-question loop across 10 ambiguity categories, with dated write-back into the plan's `## Clarifications` section. ## 10 categories (taxonomy verbatim from clarify.md:L308-L363) 1. **Functional Scope & Behavior** — what the system DOES. 2. **Domain & Data Model** — entities, relationships, lifecycle. 3. **Interaction & UX Flow** — surfaces, sequencing, defaults. 4. **Non-Functional Quality Attributes** — perf, reliability, scale. 5. **Integration & External Dependencies** — upstream/downstream contracts. 6. **Edge Cases & Failure Handling** — error paths, retries, partial state. 7. **Constraints & Tradeoffs** — non-negotiables + explicit deferrals. 8. **Terminology & Consistency** — vocabulary, naming, definitions. 9. **Completion Signals** — done criteria, acceptance metrics. 10. **Misc / Placeholders** — TBDs, owner-attestation gaps, stale TODOs. Codex R1 originally missed the 10th category; R2 confirmed all 10 via `clarify.md:L308-L363`. **DO NOT** truncate to 9 categories. ## Sequential questioning loop - **Max 5 questions per session** (hard cap per `clarify....