Overhaul the scribe skill for spawn_agent briefs and P/A/I routing

plugins/corbits-skills/skills/scribe/SKILL.md

@@ -1,14 +1,54 @@
 ---
 name: scribe
-description: Skywalker spawn recipe — shakespeare writes PRODUCT.md, ARCHITECTURE.md, and IMPLEMENTATION.md.
+description: Skywalker spawn recipe — shakespeare maintains PRODUCT.md, ARCHITECTURE.md, and IMPLEMENTATION.md (P/A/I).
+argument-hint: "[product | architecture | implementation | <doc notes>]"
 ---
 
 # Scribe
 
-You are Skywalker. This skill is a spawn recipe. Spawn `task(agent="shakespeare")` for PRODUCT.md, ARCHITECTURE.md, and IMPLEMENTATION.md unless the ask is a one-line fix (DIY with write_file/edit_file).
+You are Skywalker. This skill is a slash command (`/scribe`) and is also loadable with `use_skill("scribe")`. It is a spawn recipe for the docs lane. You do not author PRODUCT.md, ARCHITECTURE.md, or IMPLEMENTATION.md yourself except a one-line fix.
 
-Spawn `task(agent="shakespeare")` with the operator args / pasted material as the brief. Shakespeare owns PRODUCT.md, ARCHITECTURE.md, and IMPLEMENTATION.md.
+Shakespeare owns the P/A/I docs. DESIGN.md is brand-reviewer, not this skill. Product code, review, and brand are out of lane — route those elsewhere.
 
-Use `ask_operator` if the doc target (P vs A vs I) is ambiguous.
+## When to DIY vs spawn
 
-Do not edit those docs yourself except a one-line fix. DESIGN.md is brand-reviewer, not this skill.
+- **One-line / typo / single-sentence clarification** in an existing P/A/I doc → DIY with `write_file` / `edit_file` on this session. Do not spawn.
+- **Anything else** (new section, new capability, cross-doc consistency, gap fill, multi-file doc update, classification judgment) → spawn shakespeare.
+
+## Fleet
+
+Prefer `spawn_agent(agent="shakespeare", …)` then `wait_agents` on the returned `agent_id`. Use `task(agent="shakespeare")` only when a single blocking spawn is enough and you need the report before anything else.
+
+## Brief
+
+Pass the operator args / pasted material as the brief. Prefer a typed spawn:
+
+- `intent` — docs maintenance for the named P/A/I target(s)
+- `success_criteria` — done-definition (which docs, what must be true when finished)
+- `do_not` — hard constraints (e.g. do not touch DESIGN.md, do not ship product source)
+- `report_focus` — what you need back (paths changed, criteria map, open Blockers)
+- `agent="shakespeare"`
+
+### Doc types (for briefing)
+
+- **PRODUCT.md** — what we build and why: user value, vision, goals, target users, business justification
+- **ARCHITECTURE.md** — how the system is structured: components, relationships, abstractions, data/control flow, technology-agnostic design
+- **IMPLEMENTATION.md** — concrete tech: libraries, protocols, formats, configuration, deployment specifics
+
+If the operator already named P vs A vs I, put that in `success_criteria`. If the ask is multi-category, tell shakespeare to decompose into distinct claims and route each — do not ask only "which document?".
+
+Use `ask_operator` when the doc target is too fuzzy to brief (not merely multi-category). Prefer concrete options grounded in the ask.
+
+## After the report
+
+Synthesize shakespeare's Summary / Findings / Blockers / Paths for the operator. Map each `success_criteria` item → pass | fail | blocked when the worker reported that way.
+
+Do not expand into product implementation, DESIGN.md / brand, or a review campaign from this skill. If the worker Blockers say the ask needs build / brand-reviewer / critique, report that and stop — do not silently re-lane inside `/scribe`.
+
+## Non-negotiables
+
+- Spawn shakespeare for substantial P/A/I work; DIY only one-line fixes.
+- Prefer `spawn_agent` + `wait_agents`; `task(agent="shakespeare")` is the single-blocking fallback.
+- Typed brief with `success_criteria` / `do_not` / `report_focus`.
+- DESIGN.md → brand-reviewer. Product code → build. Review → `/review`.
+- Do not invent architecture campaigns after criteria are met.

tests/unit/corbits-skills-catalog.test.ts

@@ -106,6 +106,22 @@ test("spawn-recipe skills contain task(agent=", async () => {
   }
 });
 
+test("scribe skill routes P/A/I docs to shakespeare via spawn_agent", async () => {
+  const skill = await Bun.file(join(pluginRoot, "skills/scribe/SKILL.md")).text();
+  expect(skill).toContain('spawn_agent(agent="shakespeare"');
+  expect(skill).toContain("wait_agents");
+  expect(skill).toContain('task(agent="shakespeare")');
+  expect(skill).toContain("PRODUCT.md");
+  expect(skill).toContain("ARCHITECTURE.md");
+  expect(skill).toContain("IMPLEMENTATION.md");
+  expect(skill).toContain("success_criteria");
+  expect(skill).toContain("brand-reviewer");
+  expect(skill).toContain("one-line");
+  expect(skill).not.toContain("max-parallel");
+  expect(skill).not.toContain("hard cap");
+  expect(skill).not.toContain("2–4");
+});
+
 test("create-issue selects Linear MCP, GitHub gh, and MEMORY.md preference", async () => {
   const skill = await Bun.file(join(pluginRoot, "skills/create-issue/SKILL.md")).text();
   expect(skill).toContain("mcp__linear__");