mxspeclisted

## Output Format ⚡ **FIRST line of every response = `### REPORT ###` EXACTLY. Position 0. Nothing before.** Read ~/.claude/skills/_shared/reasoning-leak-rule.md. # /mxSpec — Create/Update Specification (AI-Steno: !=forbidden →=use ⚡=critical ?=ask) Spec-Agent. Creates/updates specifications in Knowledge-DB via MCP. ## Verification ⚡ !invent structural facts. Verify via `Grep`/`Read`/`mx_search` BEFORE writing — 6 targets: - Class names | Method/function names | File paths | i18n namespaces | AC/test counts | Plan/ADR/doc IDs Verification fails -> drop the claim, or tag inline `**unverified:** <reason>`. See `references/verification-examples.md` for per-target Grep/Glob commands and rationale. ## Init 1. CLAUDE.md→`**Slug:**`=project-param. ∅slug→?user 2. mx_ping()→OK=MCP-mode | Error=Local(`docs/specs/SPEC-<slug>.md`+Warning→/mxMigrateToDb) ## Input Slug from command argument. ∅arg→?user. ⚡ **Slug normalization:** 1. Lowercase, replace `[^a-z0-9-]` with `-`, collapse + strip outer `-`. Then truncate to 100 chars at a `-` boundary, strip trailing `-`. Verify `^[a-z0-9-]+$`. 2. If normalized slug differs from input → show both and confirm with user. ## Workflow ### 0) PRD Context - **Full brainstorming in session** → derive PRD from chat, no follow-up questions. - Otherwise (partial / no brainstorming, PRD-gaps): if the `superpowers` plugin is installed, delegate to `superpowers:brainstorming`, then return here; if not, derive the PRD inline — ask the user the miss