superpowers-mcp 6.4.4 → 6.4.5

package/README.es.md

@@ -2,7 +2,7 @@
 
 [English](README.md) | [繁體中文](README.zh-TW.md) | [日本語](README.ja.md) | [한국어](README.ko.md) | [Español](README.es.md) | [Português (BR)](README.pt-BR.md) | [हिन्दी](README.hi.md)
 
-[![Versión](https://img.shields.io/badge/version-6.4.4-blue.svg)](https://github.com/Poseidoncode/superpowers-mcp)
+[![Versión](https://img.shields.io/badge/version-6.4.5-blue.svg)](https://github.com/Poseidoncode/superpowers-mcp)
 [![Licencia](https://img.shields.io/badge/license-MIT-green.svg)](LICENSE)
 
 Este documento resume la información y las instrucciones de uso para empaquetar las habilidades (skills) de Superpowers y el sistema de flujos de trabajo autónomos en un servidor **Model Context Protocol (MCP)** independiente, de alto rendimiento y seguro.
@@ -162,7 +162,7 @@ Para ayudarte a elegir el skill adecuado, hemos estructurado los 15 skills a lo
 | # | Fase del SDLC | Nombre del skill | Qué hace (propósito y valor principal) | Escenario recomendado |
 | :-: | :--- | :--- | :--- | :--- |
 | 1 | **🚀 Planificación y diseño** | **`brainstorming`** | **Requisitos y diseño de arquitectura**: Explora opciones y restricciones antes de programar; produce especificaciones de diseño; incluye revisión de IU en el navegador con Visual Companion. | Antes de empezar cualquier funcionalidad nueva o cambio importante; evita saltar directamente al código. |
-| 2 | **🚀 Planificación y diseño** | **`writing-plans`** | **Planificación de la implementación**: Descompone las especificaciones en tareas pequeñas y verificables, con skills recomendados y contratos de archivos. | Antes de refactorizaciones multifichero, migraciones complejas o implementaciones grandes. |
+| 2 | **🚀 Planificación y diseño** | **`writing-plans`** | **Planificación de la implementación**: Descompone las especificaciones en tareas pequeñas y verificables, con skills recomendados y referencias exactas de archivos. | Antes de refactorizaciones multifichero, migraciones complejas o implementaciones grandes. |
 | 3 | **💻 Implementación** | **`executing-plans`** | **Ejecución del plan en la sesión**: Ejecuta cada tarea paso a paso en la sesión actual y luego hace una revisión de toda la rama al final. | Ejecución de planes por lotes dentro de la misma sesión sin crear subagentes. |
 | 4 | **💻 Implementación** | **`subagent-driven-development`** | **Desarrollo dirigido por subagentes (SDD)**: Despacha subagentes nuevos y aislados por tarea, con revisiones adversariales de doble capa. | Modelo de ejecución recomendado para planes complejos, sin contaminación de contexto. |
 | 5 | **💻 Implementación** | **`test-driven-development`** | **Desarrollo guiado por pruebas (TDD)**: Aplica ciclos estrictos de Rojo ➔ Verde ➔ Refactorización para una cobertura sólida. | Al implementar funcionalidades lógicamente difíciles o algoritmos críticos. |
@@ -179,7 +179,14 @@ Para ayudarte a elegir el skill adecuado, hemos estructurado los 15 skills a lo
 
 ## 🆕 Novedades recientes
 
-### v6.4.4 (versión actual)
+### v6.4.5 (versión actual)
+
+- **Sincronización con `obra/superpowers@8ca22db` (v6.4.2, 2026-09-25, PR #2384 "leaner plans")**: `writing-plans` ahora registra las decisiones que necesita quien implementa (firmas, aserciones de prueba, valores del spec) en lugar de transcribir código: nuevo encabezado de plan centrado en el spec con `Spec:` explícito, nueva plantilla `## What a Step Contains`, `## Bite-Sized Task Granularity` renombrado a `## Step Granularity` y `## Self-Review` con siete comprobaciones.
+- **Contenido del fork preservado en el merge**: la sección propia `## Two Plan Shapes` y `skeleton-first-plans.md` se mantienen intactos; `plan-document-reviewer-prompt.md` permanece en el fork porque `src/server.ts` lo renderiza como prompt MCP `plan-reviewer`.
+- **Baseline de drift rehecha en `8ca22db`** (74/74 archivos de skills upstream, `npm run drift` limpio); nuevo test de regresión `Test 19 (v6.4.2)` en `tests/upstream_sync_test.js`.
+- **Documentación e higiene del escaneo**: la matriz de skills de este README y los ejemplos de `docs/skill-compositions.*` coinciden con la plantilla actual de `writing-plans` (referencias exactas de archivos, no los "contratos de archivos" ya eliminados); `npm run drift` ignora archivos del sistema (`.DS_Store`).
+
+### v6.4.4
 
 - **Corrección CodeQL y parche de seguridad (2026-09-28)**:
   - El análisis de código queda en **0 abiertos / 7 corregidos**: se cerraron las 3 alertas reportadas contra v6.4.3 en `src/setup-runner.ts` (detalles en [SECURITY.md](SECURITY.md)).

package/README.hi.md

@@ -2,7 +2,7 @@
 
 [English](README.md) | [繁體中文](README.zh-TW.md) | [日本語](README.ja.md) | [한국어](README.ko.md) | [Español](README.es.md) | [Português (BR)](README.pt-BR.md) | [हिन्दी](README.hi.md)
 
-[![संस्करण](https://img.shields.io/badge/version-6.4.4-blue.svg)](https://github.com/Poseidoncode/superpowers-mcp)
+[![संस्करण](https://img.shields.io/badge/version-6.4.5-blue.svg)](https://github.com/Poseidoncode/superpowers-mcp)
 [![लाइसेंस](https://img.shields.io/badge/license-MIT-green.svg)](LICENSE)
 
 यह दस्तावेज़ Superpowers skills लाइब्रेरी और स्वायत्त वर्कफ़्लो सिस्टम को एक स्वतंत्र, उच्च-प्रदर्शन और सुरक्षित **Model Context Protocol (MCP)** सर्वर के रूप में पैकेज करने की जानकारी और उपयोग निर्देशों का सारांश है।
@@ -162,7 +162,7 @@ systematic-debugging ➔ using-git-worktrees ➔ dispatching-parallel-agents ➔
 | # | SDLC चरण | Skill नाम | यह क्या करता है (उद्देश्य व मुख्य मूल्य) | अनुशंसित परिदृश्य |
 | :-: | :--- | :--- | :--- | :--- |
 | 1 | **🚀 योजना व डिज़ाइन** | **`brainstorming`** | **आवश्यकता व आर्किटेक्चर डिज़ाइन**: कोडिंग से पहले विकल्पों और बाधाओं का पता लगाता है; डिज़ाइन Spec आउटपुट करता है; Visual Companion ब्राउज़र UI समीक्षा सहित। | किसी भी नई सुविधा या बड़े बदलाव से पहले; सीधे कोड में कूदने से रोकता है। |
-| 2 | **🚀 योजना व डिज़ाइन** | **`writing-plans`** | **कार्यान्वयन योजना**: Spec को छोटे, परीक्षणयोग्य कार्यों में बाँटता है, अनुशंसित Skills और फ़ाइल अनुबंधों सहित। | मल्टी-फ़ाइल रीफ़ैक्टर, जटिल माइग्रेशन या बड़े कार्यान्वयन से पहले। |
+| 2 | **🚀 योजना व डिज़ाइन** | **`writing-plans`** | **कार्यान्वयन योजना**: Spec को छोटे, परीक्षणयोग्य कार्यों में बाँटता है, अनुशंसित Skills और सटीक फ़ाइल संदर्भों सहित। | मल्टी-फ़ाइल रीफ़ैक्टर, जटिल माइग्रेशन या बड़े कार्यान्वयन से पहले। |
 | 3 | **💻 कार्यान्वयन** | **`executing-plans`** | **सत्र में योजना निष्पादन**: वर्तमान सत्र में हर नियोजित कार्य चरण-दर-चरण चलाता है, फिर अंत में पूरे ब्रांच की एक समीक्षा करता है। | बिना subagents बनाए उसी सत्र में योजनाओं का बैच निष्पादन। |
 | 4 | **💻 कार्यान्वयन** | **`subagent-driven-development`** | **Subagent-चालित विकास (SDD)**: प्रति कार्य ताज़ा, संदर्भ-पृथक subagents भेजता है, दोहरी प्रतिकूल समीक्षाओं सहित। | जटिल योजनाओं के लिए अनुशंसित निष्पादन मॉडल, संदर्भ प्रदूषण समाप्त करता है। |
 | 5 | **💻 कार्यान्वयन** | **`test-driven-development`** | **टेस्ट-चालित विकास (TDD)**: सख्त Red ➔ Green ➔ Refactor चक्र लागू करता है, मज़बूत टेस्ट कवरेज सुनिश्चित करता है। | तार्किक रूप से चुनौतीपूर्ण सुविधाओं या महत्वपूर्ण एल्गोरिदम लागू करते समय। |
@@ -179,7 +179,14 @@ systematic-debugging ➔ using-git-worktrees ➔ dispatching-parallel-agents ➔
 
 ## 🆕 हाल के अपडेट
 
-### v6.4.4 (नवीनतम)
+### v6.4.5 (नवीनतम)
+
+- **अपस्ट्रीम `obra/superpowers@8ca22db` (v6.4.2, 2026-09-25, PR #2384 "leaner plans") के साथ सिंक**: `writing-plans` अब कोड की प्रतिलिपि के बजाय उन निर्णयों को दर्ज करता है जिन्हें लागू करने वाले को चाहिए (हस्ताक्षर, टेस्ट असर्टन, spec मान) — स्पष्ट `Spec:` पथ वाला नया spec-केंद्रित प्लान हेडर, नया `## What a Step Contains` चरण टेम्पलेट, `## Bite-Sized Task Granularity` से `## Step Granularity` नाम परिवर्तन, और सात-जाँच वाला `## Self-Review`।
+- **फोर्क सामग्री मर्ज में सुरक्षित**: फोर्क-केवल `## Two Plan Shapes` अनुभाग और `skeleton-first-plans.md` अक्षुण्ण रहते हैं; `plan-document-reviewer-prompt.md` फोर्क में रहता है क्योंकि `src/server.ts` उसे `plan-reviewer` MCP प्रॉम्प्ट के रूप में रेंडर करता है।
+- **drift baseline `8ca22db` पर फिर से दर्ज** (74/74 अपस्ट्रीम स्किल फ़ाइलें, `npm run drift` साफ़); `tests/upstream_sync_test.js` में नया रिग्रेशन `Test 19 (v6.4.2)`।
+- **दस्तावेज़ी और स्कैन स्वच्छता**: इस README की स्किल सारणी और `docs/skill-compositions.*` उदाहरण वर्तमान `writing-plans` टेम्पलेट से मेल खाते हैं (हटाए गए "फ़ाइल अनुबंध" के बजाय सटीक फ़ाइल संदर्भ); `npm run drift` अब OS फ़ाइलों (`.DS_Store`) को फोर्क-केवल नहीं गिनता।
+
+### v6.4.4
 
 - **CodeQL सुधार व सुरक्षा पैच (2026-09-28)**:
   - कोड स्कैनिंग अब **0 खुले / 7 ठीक**: v6.4.3 के विरुद्ध रिपोर्ट किए गए `src/setup-runner.ts` के 3 अलर्ट बंद (विवरण [SECURITY.md](SECURITY.md) में)।

package/README.ja.md

@@ -2,7 +2,7 @@
 
 [English](README.md) | [繁體中文](README.zh-TW.md) | [日本語](README.ja.md) | [한국어](README.ko.md) | [Español](README.es.md) | [Português (BR)](README.pt-BR.md) | [हिन्दी](README.hi.md)
 
-[![バージョン](https://img.shields.io/badge/version-6.4.4-blue.svg)](https://github.com/Poseidoncode/superpowers-mcp)
+[![バージョン](https://img.shields.io/badge/version-6.4.5-blue.svg)](https://github.com/Poseidoncode/superpowers-mcp)
 [![License](https://img.shields.io/badge/license-MIT-green.svg)](LICENSE)
 
 このドキュメントは、Superpowers スキルライブラリと自律型ワークフローを、独立した高パフォーマンスかつ安全な **Model Context Protocol (MCP)** サーバーにパッケージ化した使用説明書です。
@@ -161,7 +161,7 @@ systematic-debugging ➔ using-git-worktrees ➔ dispatching-parallel-agents ➔
 | # | 開発フェーズ (Phase) | スキル名 (Skill Name) | 役割とコアバリュー (Purpose & Core Value) | 推奨利用シナリオ (Recommended Scenario) |
 | :-: | :--- | :--- | :--- | :--- |
 | 1 | **🚀 計画と設計** | **`brainstorming`** | **要件定義と設計探索**：コードを書く前に設計案や制約を明確化し仕様書を作成。Visual Companion による画面レビューも提供。 | 新機能や大幅な改修の開始前。AI がいきなりコードを書き始めるのを防止。 |
-| 2 | **🚀 計画と設計** | **`writing-plans`** | **実装計画の作成**：仕様書を独立検証可能なタスク一覧に分解し、Recommended Skill と変更対象を明記。 | 複数ファイルのリファクタリングや複雑な移行作業の前に実行計画を確立。 |
+| 2 | **🚀 計画と設計** | **`writing-plans`** | **実装計画の作成**：仕様書を独立検証可能なタスク一覧に分解し、Recommended Skill と変更対象ファイルの参照を明記。 | 複数ファイルのリファクタリングや複雑な移行作業の前に実行計画を確立。 |
 | 3 | **💻 実装と開発** | **`executing-plans`** | **計画の順次実行**：現在のセッションで全タスクをステップバイステップで実行し、最後にブランチ全体を一度レビュー。 | サブエージェントを起動せず、同一セッション内で計画を順次実行したい時。 |
 | 4 | **💻 実装と開発** | **`subagent-driven-development`** | **サブエージェント駆動開発 (SDD)**：タスクごとにクリーンなコンテキストのサブエージェントを起動し、2 段階の対抗的レビューを実施。 | 複雑な計画を実行する際の推奨方式。コンテキスト汚染を防ぎ精度を向上。 |
 | 5 | **💻 実装と開発** | **`test-driven-development`** | **テスト駆動開発 (TDD)**：Red ➔ Green ➔ Refactor サイクルを厳格に適用し、テストを伴う高品質なコードを実装。 | ロジックの複雑な機能やコアアルゴリズムの実装時。 |
@@ -180,7 +180,14 @@ systematic-debugging ➔ using-git-worktrees ➔ dispatching-parallel-agents ➔
 
 ## 🆕 最近の更新
 
-### v6.4.4（最新）
+### v6.4.5（最新）
+
+- **上流 `obra/superpowers@8ca22db`（上流 v6.4.2、2026-09-25、PR #2384「leaner plans」）へ同期**: `writing-plans` はコードの書き写しではなく、実装者が必要とする判断（シグネチャ、テストのアサーション、spec の値）を記録します。明示的な `Spec:` 路を持つ spec 優先の新プランヘッダ、新セクション `## What a Step Contains`、`## Bite-Sized Task Granularity` から `## Step Granularity` への改名、7 項目チェックの `## Self-Review`。
+- **フォーク固有の内容を保持**: フォーク独自の `## Two Plan Shapes` と `skeleton-first-plans.md` はマージ後も維持されます。`plan-document-reviewer-prompt.md` は `src/server.ts` が `plan-reviewer` MCP プロンプトとして使うため、フォークに残します。
+- **drift baseline を `8ca22db` で再記録**（上流スキルファイル 74/74、`npm run drift` クリーン）。`tests/upstream_sync_test.js` に回帰テスト `Test 19 (v6.4.2)` を追加。
+- **ドキュメントとスキャンの衛生**: この README のスキル一覧と `docs/skill-compositions.*` の例が現在の `writing-plans` テンプレートに一致（削除済みの「ファイル契約」ではなく正確なファイル参照）；`npm run drift` は OS ファイル（`.DS_Store`）を fork 専用ファイルとして列挙しなくなった。
+
+### v6.4.4
 
 - **CodeQL 是正とセキュリティパッチ（2026-09-28）**：
   - コードスキャンは **未対応 0 件／修正済み 7 件** に：v6.4.3 で報告された `src/setup-runner.ts` の 3 件のアラートを解消（詳細は [SECURITY.md](SECURITY.md)）。

package/README.ko.md

@@ -2,7 +2,7 @@
 
 [English](README.md) | [繁體中文](README.zh-TW.md) | [日本語](README.ja.md) | [한국어](README.ko.md) | [Español](README.es.md) | [Português (BR)](README.pt-BR.md) | [हिन्दी](README.hi.md)
 
-[![Version](https://img.shields.io/badge/version-6.4.4-blue.svg)](https://github.com/Poseidoncode/superpowers-mcp)
+[![Version](https://img.shields.io/badge/version-6.4.5-blue.svg)](https://github.com/Poseidoncode/superpowers-mcp)
 [![License](https://img.shields.io/badge/license-MIT-green.svg)](LICENSE)
 
 이 문서는 Superpowers 스킬 라이브러리와 자율 에이전트 워크플로우를 독립적이고 고성능이며 안전한 **Model Context Protocol (MCP)** 서버로 패키징한 사용 지침을 요약한 것입니다.
@@ -161,7 +161,7 @@ systematic-debugging ➔ using-git-worktrees ➔ dispatching-parallel-agents ➔
 | # | 개발 단계 (Phase) | 스킬 이름 (Skill Name) | 역할 및 핵심 가치 (Purpose & Core Value) | 추천 사용 시나리오 (Recommended Scenario) |
 | :-: | :--- | :--- | :--- | :--- |
 | 1 | **🚀 계획 및 설계** | **`brainstorming`** | **요구사항 명확화 및 설계 탐색**：코드 작성 전 아키텍처 방안과 제약을 명확히 하고 스펙을 산출. Visual Companion 브라우저 UI 검토 지원. | 새 기능이나 대규모 변경을 시작하기 전, AI가 바로 코딩하는 것을 방지. |
-| 2 | **🚀 계획 및 설계** | **`writing-plans`** | **구현 계획 분해**：스펙을 독립 검증 가능한 작업 목록으로 분해하고 Recommended Skill 및 파일 계약 명시. | 여러 파일 리팩토링이나 복잡한 마이그레이션 전 명확한 청사진 수립. |
+| 2 | **🚀 계획 및 설계** | **`writing-plans`** | **구현 계획 분해**：스펙을 독립 검증 가능한 작업 목록으로 분해하고 Recommended Skill 및 정확한 파일 참조 명시. | 여러 파일 리팩토링이나 복잡한 마이그레이션 전 명확한 청사진 수립. |
 | 3 | **💻 구현 및 개발** | **`executing-plans`** | **세션 내 계획 순차 실행**：현재 세션에서 모든 작업을 단계별로 실행한 뒤 마지막에 브랜치 전체를 한 번 검토. | 서브에이전트를 생성하지 않고 동일 세션 내에서 순차적으로 계획을 실행할 때. |
 | 4 | **💻 구현 및 개발** | **`subagent-driven-development`** | **서브에이전트 주도 개발 (SDD)**：작업별로 깨끗한 컨텍스트의 서브에이전트를 디스패치하고 이중 대립 코드 리뷰 수행. | 복잡한 계획 실행 시 컨텍스트 오염을 방지하고 정확도를 높이는 권장 방식. |
 | 5 | **💻 구현 및 개발** | **`test-driven-development`** | **테스트 주도 개발 (TDD)**：Red ➔ Green ➔ Refactor 주기를 엄격히 준수하여 견고한 테스트를 갖춘 코드 구현. | 논리적으로 복잡한 기능이나 핵심 알고리즘을 구현할 때. |
@@ -178,7 +178,14 @@ systematic-debugging ➔ using-git-worktrees ➔ dispatching-parallel-agents ➔
 
 ## 🆕 최근 업데이트
 
-### v6.4.4 (최신)
+### v6.4.5 (최신)
+
+- **상류 `obra/superpowers@8ca22db` (상류 v6.4.2, 2026-09-25, PR #2384 "leaner plans") 동기화**: `writing-plans`가 코드를 옮겨 적는 대신 구현에 필요한 결정(시그니처, 테스트 단언, spec 값)을 기록합니다. 명시적 `Spec:` 경로를 가진 spec 우선 플랜 헤더, 새 `## What a Step Contains` 단계 템플릿, `## Bite-Sized Task Granularity` → `## Step Granularity` 개명, 7단계 `## Self-Review`.
+- **포크 콘텐츠 보존**: 포크 전용 `## Two Plan Shapes` 섹션과 `skeleton-first-plans.md`는 병합 후에도 유지됩니다. `plan-document-reviewer-prompt.md`는 `src/server.ts`가 `plan-reviewer` MCP 프롬프트로 렌더링하므로 포크에 유지합니다.
+- **drift baseline을 `8ca22db`로 재기록** (상류 스킬 파일 74/74, `npm run drift` 클린). `tests/upstream_sync_test.js`에 회귀 테스트 `Test 19 (v6.4.2)` 추가.
+- **문서 및 스캔 위생**: 이 README의 스킬 표와 `docs/skill-compositions.*` 예제가 현재 `writing-plans` 템플릿과 일치(이미 제거된 "파일 계약" 대신 정확한 파일 참조); `npm run drift`는 OS 파일(`.DS_Store`)을 포크 전용으로 나열하지 않습니다.
+
+### v6.4.4
 
 - **CodeQL 수정 및 보안 패치 (2026-09-28)**:
   - 코드 스캔 **미해결 0건 / 수정 7건**: v6.4.3에서 보고된 `src/setup-runner.ts` 3건 알림 해소(자세한 내용은 [SECURITY.md](SECURITY.md)).

package/README.md

@@ -2,7 +2,7 @@
 
 [English](README.md) | [繁體中文](README.zh-TW.md) | [日本語](README.ja.md) | [한국어](README.ko.md) | [Español](README.es.md) | [Português (BR)](README.pt-BR.md) | [हिन्दी](README.hi.md)
 
-[![Version](https://img.shields.io/badge/version-6.4.4-blue.svg)](https://github.com/Poseidoncode/superpowers-mcp)
+[![Version](https://img.shields.io/badge/version-6.4.5-blue.svg)](https://github.com/Poseidoncode/superpowers-mcp)
 [![License](https://img.shields.io/badge/license-MIT-green.svg)](LICENSE)
 
 This document summarizes the information and usage instructions for packaging the Superpowers skills and autonomous workflow system into an independent, high-performance, and secure **Model Context Protocol (MCP)** server.
@@ -162,7 +162,7 @@ To help you choose the right skill, we have structured all 15 skills across the
 | # | SDLC Phase | Skill Name | What It Does (Purpose & Core Value) | Recommended Scenario |
 | :-: | :--- | :--- | :--- | :--- |
 | 1 | **🚀 Planning & Design** | **`brainstorming`** | **Requirements & Architecture Design**: Explores options and constraints before coding; outputs Design Specs; includes Visual Companion browser UI review. | Before starting any new feature or major change; prevents jumping straight into code. |
-| 2 | **🚀 Planning & Design** | **`writing-plans`** | **Implementation Planning**: Decomposes specs into bite-sized, testable tasks annotated with Recommended Skills and file contracts. | Before multi-file refactoring, complex migrations, or major implementations. |
+| 2 | **🚀 Planning & Design** | **`writing-plans`** | **Implementation Planning**: Decomposes specs into bite-sized, testable tasks annotated with Recommended Skills and exact file references. | Before multi-file refactoring, complex migrations, or major implementations. |
 | 3 | **💻 Implementation** | **`executing-plans`** | **In-Session Plan Execution**: Executes every planned task step-by-step in the current session, then one whole-branch review at the end. | Batch execution of plans within the same session without spawning subagents. |
 | 4 | **💻 Implementation** | **`subagent-driven-development`** | **Subagent-Driven Development (SDD)**: Dispatches fresh, context-isolated subagents per task with dual-layer adversarial reviews. | Recommended execution model for complex plans to eliminate context pollution. |
 | 5 | **💻 Implementation** | **`test-driven-development`** | **Test-Driven Development (TDD)**: Enforces strict Red ➔ Green ➔ Refactor cycles ensuring robust test coverage. | When implementing logically challenging features or critical algorithms. |
@@ -179,7 +179,14 @@ To help you choose the right skill, we have structured all 15 skills across the
 
 ## 🆕 Recent Updates
 
-### v6.4.4 (Latest)
+### v6.4.5 (Latest)
+
+- **Upstream sync to `obra/superpowers@8ca22db` (upstream v6.4.2, 2026-09-25, PR #2384 "leaner plans")**: `writing-plans` now records the decisions an implementer needs (signatures, test assertions, spec values) instead of transcribing code — new spec-first plan header with an explicit `Spec:` path, new `## What a Step Contains` step template, `## Bite-Sized Task Granularity` renamed `## Step Granularity`, and a seven-check `## Self-Review`.
+- **Fork content preserved through the three-way merge**: the fork-only `## Two Plan Shapes` section and `skeleton-first-plans.md` survive intact; `plan-document-reviewer-prompt.md` stays in the fork because `src/server.ts` renders it as the `plan-reviewer` MCP prompt.
+- **Drift baseline re-recorded** at `8ca22db` (74/74 upstream skill files, `npm run drift` clean); new regression `Test 19 (v6.4.2)` in `tests/upstream_sync_test.js`.
+- **Docs & scan hygiene**: the README skill matrix and `docs/skill-compositions.*` examples now match the current `writing-plans` template (exact file references, not the long-removed "file contracts"); `npm run drift` ignores OS junk (`.DS_Store`) instead of listing it as a fork-only file.
+
+### v6.4.4
 
 - **CodeQL remediation & security patch (2026-09-28)**:
   - Code scanning is now **0 open / 7 fixed**: closed the 3 alerts reported against v6.4.3 in `src/setup-runner.ts` (details in [SECURITY.md](SECURITY.md)).

package/README.pt-BR.md

@@ -2,7 +2,7 @@
 
 [English](README.md) | [繁體中文](README.zh-TW.md) | [日本語](README.ja.md) | [한국어](README.ko.md) | [Español](README.es.md) | [Português (BR)](README.pt-BR.md) | [हिन्दी](README.hi.md)
 
-[![Versão](https://img.shields.io/badge/version-6.4.4-blue.svg)](https://github.com/Poseidoncode/superpowers-mcp)
+[![Versão](https://img.shields.io/badge/version-6.4.5-blue.svg)](https://github.com/Poseidoncode/superpowers-mcp)
 [![Licença](https://img.shields.io/badge/license-MIT-green.svg)](LICENSE)
 
 Este documento resume as informações e instruções de uso para empacotar as skills do Superpowers e o sistema de fluxos de trabalho autônomos em um servidor **Model Context Protocol (MCP)** independente, de alta performance e seguro.
@@ -162,7 +162,7 @@ Para ajudar você a escolher a skill certa, estruturamos as 15 skills ao longo d
 | # | Fase do SDLC | Nome da skill | O que faz (propósito e valor principal) | Cenário recomendado |
 | :-: | :--- | :--- | :--- | :--- |
 | 1 | **🚀 Planejamento e design** | **`brainstorming`** | **Requisitos e design de arquitetura**: Explora opções e restrições antes de codar; gera specs de design; inclui revisão de UI no navegador com Visual Companion. | Antes de começar qualquer feature nova ou grande mudança; evita pular direto para o código. |
-| 2 | **🚀 Planejamento e design** | **`writing-plans`** | **Planejamento da implementação**: Decompõe specs em tarefas pequenas e testáveis, com skills recomendadas e contratos de arquivos. | Antes de refatorações multiarquivo, migrações complexas ou implementações grandes. |
+| 2 | **🚀 Planejamento e design** | **`writing-plans`** | **Planejamento da implementação**: Decompõe specs em tarefas pequenas e testáveis, com skills recomendadas e referências exatas de arquivos. | Antes de refatorações multiarquivo, migrações complexas ou implementações grandes. |
 | 3 | **💻 Implementação** | **`executing-plans`** | **Execução do plano na sessão**: Executa cada tarefa passo a passo na sessão atual e depois faz uma revisão do branch inteiro no final. | Execução de planos em lote dentro da mesma sessão, sem criar subagentes. |
 | 4 | **💻 Implementação** | **`subagent-driven-development`** | **Desenvolvimento dirigido por subagentes (SDD)**: Despacha subagentes novos e isolados por tarefa, com revisões adversariais em duas camadas. | Modelo de execução recomendado para planos complexos, sem poluição de contexto. |
 | 5 | **💻 Implementação** | **`test-driven-development`** | **Desenvolvimento guiado por testes (TDD)**: Aplica ciclos rigorosos de Vermelho ➔ Verde ➔ Refatoração, garantindo cobertura robusta. | Ao implementar features logicamente desafiadoras ou algoritmos críticos. |
@@ -179,7 +179,14 @@ Para ajudar você a escolher a skill certa, estruturamos as 15 skills ao longo d
 
 ## 🆕 Novidades recentes
 
-### v6.4.4 (atual)
+### v6.4.5 (atual)
+
+- **Sincronização com `obra/superpowers@8ca22db` (v6.4.2, 2026-09-25, PR #2384 "leaner plans")**: `writing-plans` agora registra as decisões que o implementador precisa (assinaturas, asserções de teste, valores do spec) em vez de transcrever código: novo cabeçalho de plano centrado no spec com `Spec:` explícito, nova plantilha `## What a Step Contains`, `## Bite-Sized Task Granularity` renomeado para `## Step Granularity` e `## Self-Review` com sete verificações.
+- **Conteúdo do fork preservado no merge**: a seção própria `## Two Plan Shapes` e `skeleton-first-plans.md` permanecem intactos; `plan-document-reviewer-prompt.md` fica no fork porque `src/server.ts` o renderiza como prompt MCP `plan-reviewer`.
+- **Baseline de drift regravada em `8ca22db`** (74/74 arquivos de skills upstream, `npm run drift` limpo); novo teste de regressão `Test 19 (v6.4.2)` em `tests/upstream_sync_test.js`.
+- **Documentação e higiene do scan**: a matriz de skills deste README e os exemplos de `docs/skill-compositions.*` seguem o template atual do `writing-plans` (referências exatas de arquivos, não os "contratos de arquivos" já removidos); `npm run drift` ignora arquivos do sistema (`.DS_Store`).
+
+### v6.4.4
 
 - **Correção CodeQL e patch de segurança (2026-09-28)**:
   - O code scanning fica em **0 abertos / 7 corrigidos**: fechados os 3 alertas reportados contra a v6.4.3 em `src/setup-runner.ts` (detalhes em [SECURITY.md](SECURITY.md)).

package/README.zh-TW.md

@@ -2,7 +2,7 @@
 
 [English](README.md) | [繁體中文](README.zh-TW.md) | [日本語](README.ja.md) | [한국어](README.ko.md) | [Español](README.es.md) | [Português (BR)](README.pt-BR.md) | [हिन्दी](README.hi.md)
 
-[![版本](https://img.shields.io/badge/version-6.4.4-blue.svg)](https://github.com/Poseidoncode/superpowers-mcp)
+[![版本](https://img.shields.io/badge/version-6.4.5-blue.svg)](https://github.com/Poseidoncode/superpowers-mcp)
 [![授權](https://img.shields.io/badge/license-MIT-green.svg)](LICENSE)
 
 本文檔總結了將 Superpowers 技能庫與自主 Agent 工作流架構打包成獨立、高效能且安全加固的 **Model Context Protocol (MCP)** 伺服器之相關資訊與使用說明。
@@ -161,7 +161,7 @@ systematic-debugging ➔ using-git-worktrees ➔ dispatching-parallel-agents ➔
 | # | 開發階段 (Phase) | 技能名稱 (Skill Name) | 它是幹嘛用的？ (Purpose & Core Value) | 推薦使用情境 (Recommended Scenario) |
 | :-: | :--- | :--- | :--- | :--- |
 | 1 | **🚀 規劃與設計** | **`brainstorming`** | **需求澄清與設計探索**：在寫代碼前探索架構方案、釐清邊界，產出 Spec。內建 Visual Companion 瀏覽器即時設計審查。 | 啟動任何新功能或大改版前，防止 AI 直接衝進去寫 code。 |
-| 2 | **🚀 規劃與設計** | **`writing-plans`** | **實作計畫拆解**：將設計規格分解為獨立可測試的原子任務清單，標註檔案契約與 Recommended Skill。 | 進行多檔案重構、複雜遷移或大型開發前，建立清晰藍圖。 |
+| 2 | **🚀 規劃與設計** | **`writing-plans`** | **實作計畫拆解**：將設計規格分解為獨立可測試的原子任務清單，標註精確檔案參照與 Recommended Skill。 | 進行多檔案重構、複雜遷移或大型開發前，建立清晰藍圖。 |
 | 3 | **💻 開發與實作** | **`executing-plans`** | **會話內計畫執行**：在當前會話中依據計畫逐步執行全部任務，完成後再做一次全分支審查。 | 不需要開多個 Subagent 時的連續計畫執行。 |
 | 4 | **💻 開發與實作** | **`subagent-driven-development`** | **子代理驅動開發 (SDD)**：為各任務派發乾淨上下文的獨立 Subagent 實作，並在任務間發起雙層對抗式代碼審查。 | 推薦的複雜計畫執行方式，防止上下文污染並提高精確度。 |
 | 5 | **💻 開發與實作** | **`test-driven-development`** | **測試驅動開發 (TDD)**：嚴格執行「紅燈（寫失敗測試）➔ 綠燈（最小實作）➔ 重構」循環。 | 實作邏輯複雜功能時，確保代碼隨附測試且具備回歸防護。 |
@@ -178,7 +178,14 @@ systematic-debugging ➔ using-git-worktrees ➔ dispatching-parallel-agents ➔
 
 ## 🆕 最近更新
 
-### v6.4.4（最新版）
+### v6.4.5（最新版）
+
+- **上游同步至 `obra/superpowers@8ca22db`（上游 v6.4.2，2026-09-25，PR #2384「leaner plans」）**：`writing-plans` 改為記錄實作時需要的決策（函式簽章、測試斷言、spec 數值），而非抄寫程式碼——新增明確 `Spec:` 路徑的 spec 優先計畫標頭、新增 `## What a Step Contains` 步驟範本、`## Bite-Sized Task Granularity` 更名為 `## Step Granularity`，以及含七項檢查的 `## Self-Review`。
+- **三方合併保留 fork 內容**：fork 獨有的 `## Two Plan Shapes` 段落與 `skeleton-first-plans.md` 完整保留；`plan-document-reviewer-prompt.md` 留在 fork，因為 `src/server.ts` 用它產生 `plan-reviewer` MCP prompt。
+- **Drift baseline 重錄於 `8ca22db`**（74/74 上游 skill 檔案，`npm run drift` 已歸零）；`tests/upstream_sync_test.js` 新增回歸測試 `Test 19 (v6.4.2)`。
+- **文件與掃描衛生**：README 技能矩陣與 `docs/skill-compositions.*` 範例已對齊目前的 `writing-plans` 範本（改用精確檔案參照，不再提到早已移除的「檔案契約」）；`npm run drift` 不再把系統雜項（`.DS_Store`）列為 fork 獨有檔案。
+
+### v6.4.4
 
 - **CodeQL 修復與安全更新（2026-09-28）**：
   - Code scanning 已 **0 未修復 / 7 已修復**：補上 v6.4.3 通報的 3 個 `src/setup-runner.ts` 警報（詳見 [SECURITY.md](SECURITY.md)）。

package/docs/skill-compositions.es.md

@@ -148,13 +148,17 @@ En planes generados por `writing-plans`, especifica los skills recomendados por
 
 ```markdown
 ### Task 1: Implement Token Authentication Middleware
-- **Goal**: Validate JWT tokens and extract user claims
-- **Target Files**: `src/auth/jwt.ts`, `tests/auth/jwt.test.ts`
-- **Recommended Skill**: `superpowers:test-driven-development`
-- **Task Brief**:
-  1. Write failing test for expired and invalid signatures (FAIL)
-  2. Implement minimal signature verification (PASS)
-  3. Refactor with strict type safety
+
+**Files:**
+- Create: `src/auth/jwt.ts`
+- Test: `tests/auth/jwt.test.ts`
+
+**Recommended Skill:** `superpowers:test-driven-development` (or relevant skill)
+
+**Checklist:**
+- [ ] 1. Write failing test for expired and invalid signatures (FAIL) — verify with: `npm test`
+- [ ] 2. Implement minimal signature verification (PASS) — verify with: `npm test`
+- [ ] 3. Refactor with strict type safety — verify with: `npx tsc --noEmit`
 ```
 
 ### Protocolo de despacho controlador → subagente

package/docs/skill-compositions.hi.md

@@ -148,13 +148,17 @@ flowchart LR
 
 ```markdown
 ### Task 1: Implement Token Authentication Middleware
-- **Goal**: Validate JWT tokens and extract user claims
-- **Target Files**: `src/auth/jwt.ts`, `tests/auth/jwt.test.ts`
-- **Recommended Skill**: `superpowers:test-driven-development`
-- **Task Brief**:
-  1. Write failing test for expired and invalid signatures (FAIL)
-  2. Implement minimal signature verification (PASS)
-  3. Refactor with strict type safety
+
+**Files:**
+- Create: `src/auth/jwt.ts`
+- Test: `tests/auth/jwt.test.ts`
+
+**Recommended Skill:** `superpowers:test-driven-development` (or relevant skill)
+
+**Checklist:**
+- [ ] 1. Write failing test for expired and invalid signatures (FAIL) — verify with: `npm test`
+- [ ] 2. Implement minimal signature verification (PASS) — verify with: `npm test`
+- [ ] 3. Refactor with strict type safety — verify with: `npx tsc --noEmit`
 ```
 
 ### नियंत्रक-से-subagent प्रेषण प्रोटोकॉल

package/docs/skill-compositions.ja.md

@@ -151,13 +151,17 @@ flowchart LR
 
 ```markdown
 ### Task 1: トークン認証ミドルウェアの実装
-- **Goal**: JWT トークンの検証とクレーム抽出
-- **Target Files**: `src/auth/jwt.ts`, `tests/auth/jwt.test.ts`
-- **Recommended Skill**: `superpowers:test-driven-development`
-- **Task Brief**:
-  1. 期限切れおよび無効な署名の失敗テストを作成 (FAIL)
-  2. 最小限の検証ロジックを実装してパスさせる (PASS)
-  3. 厳格な型安全性を確保してリファクタリング
+
+**Files:**
+- Create: `src/auth/jwt.ts`
+- Test: `tests/auth/jwt.test.ts`
+
+**Recommended Skill:** `superpowers:test-driven-development` (or relevant skill)
+
+**Checklist:**
+- [ ] 1. 期限切れおよび無効な署名の失敗テストを作成 (FAIL) — verify with: `npm test`
+- [ ] 2. 最小限の検証ロジックを実装してパスさせる (PASS) — verify with: `npm test`
+- [ ] 3. 厳格な型安全性を確保してリファクタリング — verify with: `npx tsc --noEmit`
 ```
 
 ### コントローラーとサブエージェントのディスパッチプロトコル

package/docs/skill-compositions.ko.md

@@ -151,13 +151,17 @@ flowchart LR
 
 ```markdown
 ### Task 1: 토큰 인증 미들웨어 구현
-- **Goal**: JWT 토큰 검증 및 클레임 추출
-- **Target Files**: `src/auth/jwt.ts`, `tests/auth/jwt.test.ts`
-- **Recommended Skill**: `superpowers:test-driven-development`
-- **Task Brief**:
-  1. 만료 및 유효하지 않은 서명에 대한 실패 테스트 작성 (FAIL)
-  2. 최소한의 검증 로직을 구현하여 테스트 통과 (PASS)
-  3. 엄격한 타입 안전성을 확보하며 리팩토링
+
+**Files:**
+- Create: `src/auth/jwt.ts`
+- Test: `tests/auth/jwt.test.ts`
+
+**Recommended Skill:** `superpowers:test-driven-development` (or relevant skill)
+
+**Checklist:**
+- [ ] 1. 만료 및 유효하지 않은 서명에 대한 실패 테스트 작성 (FAIL) — verify with: `npm test`
+- [ ] 2. 최소한의 검증 로직을 구현하여 테스트 통과 (PASS) — verify with: `npm test`
+- [ ] 3. 엄격한 타입 안전성을 확보하며 리팩토링 — verify with: `npx tsc --noEmit`
 ```
 
 ### 컨트롤러와 서브에이전트 디스패치 프로토콜

package/docs/skill-compositions.md

@@ -148,13 +148,17 @@ In plans generated by `writing-plans`, specify recommended skills for each task:
 
 ```markdown
 ### Task 1: Implement Token Authentication Middleware
-- **Goal**: Validate JWT tokens and extract user claims
-- **Target Files**: `src/auth/jwt.ts`, `tests/auth/jwt.test.ts`
-- **Recommended Skill**: `superpowers:test-driven-development`
-- **Task Brief**:
-  1. Write failing test for expired and invalid signatures (FAIL)
-  2. Implement minimal signature verification (PASS)
-  3. Refactor with strict type safety
+
+**Files:**
+- Create: `src/auth/jwt.ts`
+- Test: `tests/auth/jwt.test.ts`
+
+**Recommended Skill:** `superpowers:test-driven-development` (or relevant skill)
+
+**Checklist:**
+- [ ] 1. Write failing test for expired and invalid signatures (FAIL) — verify with: `npm test`
+- [ ] 2. Implement minimal signature verification (PASS) — verify with: `npm test`
+- [ ] 3. Refactor with strict type safety — verify with: `npx tsc --noEmit`
 ```
 
 ### Controller-to-Subagent Dispatch Protocol

package/docs/skill-compositions.pt-BR.md

@@ -148,13 +148,17 @@ Em planos gerados por `writing-plans`, especifique as skills recomendadas por ta
 
 ```markdown
 ### Task 1: Implement Token Authentication Middleware
-- **Goal**: Validate JWT tokens and extract user claims
-- **Target Files**: `src/auth/jwt.ts`, `tests/auth/jwt.test.ts`
-- **Recommended Skill**: `superpowers:test-driven-development`
-- **Task Brief**:
-  1. Write failing test for expired and invalid signatures (FAIL)
-  2. Implement minimal signature verification (PASS)
-  3. Refactor with strict type safety
+
+**Files:**
+- Create: `src/auth/jwt.ts`
+- Test: `tests/auth/jwt.test.ts`
+
+**Recommended Skill:** `superpowers:test-driven-development` (or relevant skill)
+
+**Checklist:**
+- [ ] 1. Write failing test for expired and invalid signatures (FAIL) — verify with: `npm test`
+- [ ] 2. Implement minimal signature verification (PASS) — verify with: `npm test`
+- [ ] 3. Refactor with strict type safety — verify with: `npx tsc --noEmit`
 ```
 
 ### Protocolo de despacho controlador → subagente

package/docs/skill-compositions.zh-TW.md

@@ -151,13 +151,17 @@ flowchart LR
 
 ```markdown
 ### Task 1: 實作 Token 驗證中介軟體 (Token Middleware)
-- **Goal**: 驗證 JWT Token 並解析 Claims
-- **Target Files**: `src/auth/jwt.ts`, `tests/auth/jwt.test.ts`
-- **Recommended Skill**: `superpowers:test-driven-development`
-- **Task Brief**:
-  1. 編寫 JWT 簽名與過期驗證測試 (FAIL)
-  2. 實作驗證邏輯使測試通過 (PASS)
-  3. 重構並確保型別安全
+
+**Files:**
+- Create: `src/auth/jwt.ts`
+- Test: `tests/auth/jwt.test.ts`
+
+**Recommended Skill:** `superpowers:test-driven-development` (or relevant skill)
+
+**Checklist:**
+- [ ] 1. 編寫 JWT 簽名與過期驗證測試 (FAIL) — verify with: `npm test`
+- [ ] 2. 實作驗證邏輯使測試通過 (PASS) — verify with: `npm test`
+- [ ] 3. 重構並確保型別安全 — verify with: `npx tsc --noEmit`
 ```
 
 ### 主 Agent 與 Subagent 調度協議