app_tunnel

Branch Worktree Guardian - Git 分支和 Worktree 管理工具。Use for: (1) 新開發需求時建立隔離分支, (2) 使用 worktree 機制避免分支衝突, (3) 驗證當前工作分支正確性, (4) 預防在錯誤分支上開發

Claude Code release notes 框架影響評估工具。比對 last-reviewed 版本篩出新版本，逐項分類（對框架有幫助 / 需評估 / 無影響 / 不適用），對採用項引導建 ANA + WRAP + spawn 落地。Use when: 執行 /release-notes 看到新版本、定期檢查 CC 更新、評估新功能對專案框架的影響時。Triggers: release notes, release-notes, CC 更新, claude code 更新, 版本更新評估, 新功能評估, 框架影響評估。

Use this skill for managing git worktrees for Ticket-based development. Triggers include: creating a worktree for a new ticket, checking worktree status, viewing all worktrees, or any mention of /worktree, worktree management, feature branches, or setting up development environment.

版本發布整合工具。Use for: (1) 發布新版本（合併到 main、打 Tag、推送）, (2) 發布前健康檢查（所有 Ticket 完成？CHANGELOG 更新？）, (3) 更新版本文件（worklog 狀態、CHANGELOG）。Use when: 準備發布版本、執行 /version-release check 確認發布前狀態、完成所有 Ticket 後要收尾時。

Composes atomic, intent-revealing, grep-friendly writing (Zettelkasten) for code comments, docs, logs, prompts, schema/ticket fields, external-analysis transformation, and long-form technical articles. Use when cognitive load and token cost matter. **Also triggers during multi-round review / batch review / 寫作 audit** — provides the keyword bank (正向陳述 / 口語修辭 / 地區用語 / 廢話前綴 / 裝飾符號 / 對讀者喊話 / 自評誇飾 / 必然性框架) and frame-specific check lists that multi-round-review reviewer agents need. Triggers: 寫註解, 寫文件, 寫日誌, 寫 prompt, 寫文章, 技術文章, 商業分析, 外部分析文章, post-mortem, 架構決策, 除錯復盤, 欄位設計, atomic, reusable, 多輪審查, multi-round review, batch review, 寫作 audit, 正向陳述, 口語修辭, 字句層 grep.

寫多篇章節後做多輪 agent reviewer audit 的標準操作流程。每輪用不同 frame 切換、跨輪 finding 互不重疊、停止訊號是 frame 涵蓋而非 finding 數遞減。Round 1-A 寫作規範 reviewer 必須同步 invoke `compositional-writing` skill 的字句層 keyword bank（正向陳述 / 口語修辭 / 地區用語 / 廢話前綴 / 裝飾符號 / 對讀者喊話 / 自評誇飾 / 必然性框架）、且命中後要做語意判定（命中是候選不是判決）。觸發詞：多輪審查、Round 1/2/3、frame 切換、跨輪審查、reviewer 規劃、何時停止 review、寫作 audit、batch review、cadence 同骨化、enumeration 不窮盡、正向陳述、self-application sweep。Trigger when reviewing multiple writings via successive rounds of agent reviewers.

Use this skill whenever the user wants to create, track, query, or manage tickets. Triggers include: creating new tickets, claiming or releasing tickets, checking ticket status or progress, completing tickets, handing off work between agents, resuming interrupted tasks, migrating tickets between versions, converting plans to tickets, or any mention of /ticket, task tracking, or ticket lifecycle operations.

WRAP 決策框架 — 認知偏誤防護、選項擴增與資料充足度閘門。用於防護自動駕駛、假選項、證據不足下的倉促決策。Use when: 被困住或連續失敗 2+、準備宣告限制性結論前、偏離核心目標、根因分析、代理人失敗歸因、提案評估、重大架構或規則決策、個人化建議（健康/醫療/金錢/法律）前、Context 沉重時。Triggers: stuck, blocked, loop, no progress, 分析, debug, 限制性解法, 個人化建議, 具體推薦。

Creates and lists framework issues on the canonical framework repo (tarrragon/claude) via gh CLI. Use when tracking a framework-level problem, error-pattern canonical reference, provenance anchor, or cross-consumer fix across projects. Triggers include: framework issue, canonical issue, 跨 consumer 修復追蹤, 框架 issue, error-pattern canonical. Do NOT use for project-local docs/work-logs tickets (use the ticket skill instead).

需求追蹤文件系統（proposals/spec/usecases）的查詢、建立、導航和管理。Use for: (1) 查詢提案、規格、用例文件, (2) 建立新提案/規格/用例（從模板）, (3) 跨文件導航（從 UC 找 spec/ticket）, (4) Domain 地圖查詢, (5) 追蹤索引管理, (6) UC 測試對應驗證, (7) 提案評估與審查。Use when: user mentions PROP-, UC-, SPEC-, 功能, 需求, feature, issue, 提案, 用例, 規格, 需求文件, 需求追蹤

初始化 SaaS repo 時的設計與選型訪談協議：定錨後先過交付形態 gate（託管平台 / 垂直 SaaS / 辦公生態自動化 / BaaS / 半託管 CMS / 自建）、自建成立才從使用者操作（BDD）推導功能與風險、依 SRP / OCP 切分 domain 與 event（DDD）、再把技術維度掛在領域骨架下逐項確認、每個維度附不可沉默跳過的防護底線、產出設計決策記錄與 scaffold 建議。Triggers: 初始化 repo, 新專案, 開新服務, SaaS 選型, 技術選型, tech stack, 要不要自建, 託管平台, Shopify, Wix, Firebase, WordPress, Apps Script, DDD, domain 切分, event 驅動, event storming, BDD, 行為情境, 使用者操作盤點, 選資料庫, 選 queue, 要不要 redis, 要不要 k8s, MVP 架構, repo scaffold, 專案起手, stack 評估, 選型訪談, 架構訪談.

Use this skill whenever the user needs to check or set up development environment, initialize project dependencies, install required tools, or verify system compatibility. Triggers include: environment initialization, setup environment, check environment, install tools, verify dependencies, environment status, platform setup, or any mention of /project-init.

5W1H Decision Framework Tool. Use for: (1) Systematic decision-making before creating todos, (2) Preventing duplicate implementation, (3) Detecting avoidance behavior, (4) Ensuring agile refactor compliance with executor/dispatcher separation

Agent Teams 協作派發指南。Use when: (1) Agent A 的發現會改變 Agent B 正在進行的工作, (2) 用戶要求使用 team/swarm, (3) 多代理人需即時協商共用介面或 API 契約。涵蓋 team 建立、Ticket-Task 橋接、teammate 入職、生命週期管理。

broken-link 偵測工具。掃描 .claude/ 目錄所有 Markdown 文件中的路徑引用，偵測失效連結。Use for: (1) 一次性掃描所有 broken links, (2) 搭配 /loop 定期監控, (3) 修改規則/方法論/代理人文件後驗證路徑完整性。Use when: user runs /broken-link-check, 或搭配 /loop 定期執行, 或發現 broken link 錯誤後。

子任務拆分與 Context 卸載工具。將可拆分的大型任務分成 N 個子 Ticket，各由 Agent 獨立執行，結論直接寫入 Ticket 不回報主線程。Use for: 批量檔案評估, 大型審查任務拆分, 任何需要讀取大量資料但結果可落地到 Ticket 的任務

Chrome Extension 實機測試與 debug 工作流，以 chrome-devtools-mcp 為核心工具。Use when: (1) 完成功能後實機驗證 / manual test / 試看看 / 跑看看 / verify feature, (2) extension debug / popup 不作動 / content script 不注入 / service worker 報錯 / background 出問題, (3) 安裝 unpacked extension / load unpacked / 載入未封裝, (4) 看 console / 看 network / 看 log / view console / inspect requests, (5) 功能更新後重新載入 extension / rebuild reload / reload extension。涵蓋 Manifest V3 service worker / content script / popup / options page 的 chrome-devtools-mcp 工具呼叫流程。不取代 Puppeteer / Playwright 自動化 E2E（CI 用），定位為開發期手動驗證與 LLM-assisted debug。

認知負擔評估與審查工具。作為決策樹、代理人、Code Review 的基本參考標準。用於: (1) 任務複雜度評估, (2) 代理人升級判斷, (3) 任務拆分建議, (4) 程式碼品質審查與熱點識別

Extracts reusable patterns from Claude Code sessions and captures knowledge as atomic memory units, then evaluates whether each memory should be upgraded to framework-shared rules, methodologies, or error-patterns. Use when session ends (Stop hook), when recording technical decisions, implementation insights, or lessons learned. Handles automatic pattern detection, structured memory capture with interconnected knowledge links, and post-write upgrade decision flow to prevent cross-project principles being trapped in single-project memory.

Designs web scraping / data extraction strategies: DOM analysis, multi-tier selectors with fallbacks, validation, transformation mapping, performance planning. Use when planning to scrape a site, extract structured data from HTML/DOM, or handle pagination / anti-scraping. Triggers: 資料提取, 抓取策略, web scraping, 選擇器設計, 爬蟲規劃. Do NOT use for implementing scraper code.

決策樹助手工具。快速評估任務複雜度，提供派發建議。Use when: (1) 任務複雜度快速評估, (2) 派發代理人建議, (3) 拆分策略建議, (4) 並行可行性評估

多方案評估決策框架。用於面臨 3+ 技術方案時的結構化評估、架構決策時的系統化分析，防止衝動決策和技術債務累積。Use for: 技術方案選擇、重大架構決策、高風險技術選型

派發策略檢討工具. Use for: (1) 失敗數超預期 30% 時的策略檢討, (2) 重複分派失敗的分析, (3) 代理人選擇錯誤的修正

Manages project documentation system including CHANGELOG, worklog, tickets, error-patterns, and todolist. Use for: (1) worklog initialization and updates, (2) todolist management, (3) version collaboration workflows, (4) documentation consistency checks

錯誤模式知識庫管理工具。Use for: (1) 查詢既有錯誤經驗和防護措施 (query), (2) 記錄新發現的錯誤模式和教訓 (add), (3) Ticket 開始前查詢歷史問題避免再犯, (4) 系統化管理錯誤學習經驗。Use when: user mentions error pattern, 錯誤模式, 教訓, 經驗記錄, 學習經驗, 防護措施, 錯誤紀錄, or needs to avoid recurring issues.

證據驅動除錯流程。Use for: (1) Bug 修復的結構化流程（重現 -> failing test -> 根因 -> 最小修復 -> 回歸防護）, (2) 防止盲目猜測修復, (3) 確保每次修復可驗證、可追溯、可累積。Use when: 收到 bug report、測試失敗需要調查、生產環境問題需要修復、任何非新功能的程式碼修正。

框架無關的前端開發協議 + Playwright 驗證 + 跨領域 Stream 操作架構：DOM topology 先於 CSS、CSS / JS 邊界辨識、Playwright 三個位置（假設 / 行為 / 互動驗證）、寫成 layout 測試、framework-managed DOM 共處、Reactive 效能盤點、a11y 三道防線、Filter × Source 層錯位 + 五策略合成（適用前端 / 後端 / 演算法 / DB）。Triggers: 寫 CSS, selector 精準, CSS layers, CSS-only vs JS, class toggle, MutationObserver, observer scope, polling, framework 共處, 外部組件客製, custom UI 邊界外, playwright 驗證, browser_evaluate, layout test, focus management, aria-live, keyboard a11y, reactive 效能, runtime cost, layout reflow, lazy loading, 前端網頁開發, vanilla, vue, react, jquery, filter × source, 層錯位, paginated source, post-filter, 自動續抓, 推進 query, 誠實進度 UX, 演算法 pipeline, middleware filter, materialized view, map-reduce.

全量掃描硬編碼中文字串並生成 i18n 修正建議。檢測所有層（Model/Service/Domain/UI）中文硬編碼，生成 ARB 鍵值建議，支援批量替換工作流程。Use for: (1) 檢查專案全部硬編碼中文（不只 UI 層）, (2) 進行大規模 i18n 修復, (3) 生成 ARB 鍵值建議, (4) i18n 技術債務評估

Use when the user wants to design, redesign, shape, critique, audit, polish, clarify, distill, harden, optimize, adapt, animate, colorize, extract, or otherwise improve a frontend interface. Covers websites, landing pages, dashboards, product UI, app shells, components, forms, settings, onboarding, and empty states. Handles UX review, visual hierarchy, information architecture, cognitive load, accessibility, performance, responsive behavior, theming, anti-patterns, typography, fonts, spacing, layout, alignment, color, motion, micro-interactions, UX copy, error states, edge cases, i18n, and reusable design systems or tokens. Also use for bland designs that need to become bolder or more delightful, loud designs that should become quieter, live browser iteration on UI elements, or ambitious visual effects that should feel technically extraordinary. Not for backend-only or non-UI tasks.

LSP 優先開發策略工具。Use for: (1) 查詢 LSP 操作指令, (2) 配置新語言 LSP 插件, (3) LSP vs MCP 工具選擇決策, (4) 自建 LSP 插件指南。Use when: 需要搜尋符號/定義/引用、想知道 LSP 和 MCP Serena 各自適合什麼場景、新增語言支援需要配置 LSP 時。

Mermaid 圖表 ASCII 渲染工具。Use for: (1) 將 Mermaid 圖表轉為純文字 ASCII, (2) 在不支援圖形的環境中顯示流程圖, (3) 文件中嵌入文字版圖表

方法論撰寫指南。Use for: (1) 撰寫新方法論, (2) 改寫冗長的舊方法論, (3) 檢查方法論品質

多視角審核/code review 工具。派發三人組（含常駐委員 linux）並行掃描程式碼品質、架構設計、重構評估。Use for: 程式碼審查, 架構評估, 重構掃描, 結論審查, 系統設計變更審查, 功能規劃審查。Use when: Phase 3b 完成後 PR 前, Phase 4 重構評估前, 重大架構決策前, ANA Ticket 結論審查, 任何分析報告產出後, 規則/Skill/方法論變更後, Wave 完成審查, 規格設計完成後

修復前強制評估系統. Use for: (1) 測試失敗自動評估, (2) 編譯錯誤分類處理, (3) 強制 Ticket 開設流程

Riverpod Provider 架構設計規範 - 確保正確的依賴注入、介面隔離和測試可行性。Use for: (1) 設計新的 ViewModel/Notifier 類別, (2) 審查 Provider 依賴注入是否正確, (3) 測試中配置 ProviderScope.overrides, (4) 發現 ref.read/watch 使用錯誤時。Use when: 程式碼涉及 Riverpod Provider、Notifier、ViewModel 設計或出現 ref 操作問題時。

從需求確認到實作的對話協議：模糊指令澄清（含篩選類）、可決定 vs 該確認的邊界、失敗 2 次的轉折、覆寫成本告知、revert/checkpoint 處理、漸進驗證、工具切換時機。Triggers: 收到模糊指令, 自決還是確認, 反覆失敗, 換思路, 覆寫成本, 先還原, 先重來, placeholder, 最小範圍, 推理失敗, playwright 切換, 開發前澄清, 需求確認, 排除障礙, 逼近答案, 依 X 篩選, 只看 X, filter 範圍, 呈現決策, 開放問, ABCDE 你選哪個, 反省題, retrospective, 下一步往哪走, 五維度, 需要我繼續嗎, 要做嗎, OK 嗎, yes/no, 二選, 確認嗎.

功能範圍確認工具. Use for: (1) 測試/實作/需求不一致時的範圍確認, (2) 功能邊界不清時的澄清, (3) 防止範圍蔓延

搜尋工具使用指南。用於：(1) 選擇正確的搜尋工具, (2) 三 MCP 工作流決策樹（cbm 概念搜尋 / codegraph 呼叫圖 / serena 型別感知), (3) rg 精確文字搜尋, (4) LSP/Serena 符號搜尋, (5) 工具安裝與故障排除

Performs security review for Flutter/Dart code. Use when implementing authentication, handling user input, working with API keys or secrets, managing permissions, storing sensitive data locally, or integrating third-party APIs. Provides Flutter-specific security checklists and patterns.

Use this skill when creating a new skill, updating an existing skill's YAML frontmatter, or reviewing skill quality. Provides the official Anthropic skill specification, frontmatter rules, description writing best practices, progressive disclosure architecture, and common pitfalls to avoid. Triggers include: creating skills, skill review, frontmatter validation, SKILL.md writing.