将用户提供的聊天导出 JSON 转换为多个 Markdown 文件，并统一为问答结构。

Use when splitting rough feature notes into overview, frontend, and backend implementation docs with codebase context and contracts.

Use when building, modifying, or reviewing frontend UI where text overflow, layout, responsiveness, accessibility, and visual polish must be verified.

Use when creating one local Git commit without pushing. Supports diff-based Conventional Commit type/scope/message generation, staged or requested change handling, segmented commit bodies for non-trivial changes, and safe commit-only workflows. Trigger for requests such as "commit this", "create a commit", "commit current changes without pushing", or local-only handoff.

Use when creating, critiquing, or iterating logo and brand-mark concepts through a designer-distillation board, including brand intake, public-source designer lenses, concept territories, visual prompts, and logo quality scoring.

审计 macOS 软件安装与磁盘占用，输出可执行的分级清理建议，并在用户确认后执行低风险清理。

Create multilingual README files by translating an existing README into target languages. Use when users ask for bilingual or multilingual README output, such as "create bilingual README", "translate README to Japanese", or "add Chinese README". Supports English, German (de), Japanese (ja), Korean (ko), and Chinese (zh-CN).

Reorganize exported browser bookmarks under one user-specified topic folder into cleaner categories, remove duplicate URLs, and generate an importable Netscape HTML file. Use when users ask to analyze a bookmarks export, extract one topic directory, regroup links and subfolders, and output a browser-importable HTML file.

Guide agents through approved TweetClaw workflows for X/Twitter automation, including tweet scraping, reply search, follower export, user lookup, media, DMs, monitors, webhooks, giveaway draws, and confirmation-gated posts or replies.

指导 Agent 使用 TweetClaw 执行已获用户批准的 X/Twitter 自动化，包括推文抓取、回复搜索、粉丝导出、用户查询、媒体、私信、监控、Webhook、抽奖，以及确认后发帖或回复。

Analyze UI images and output layout description with functional analysis. Triggers when user sends a UI screenshot and says "analyze this UI", "describe this interface layout", "what does this UI do", or "output UI layout and function description".

Extract visual styles from WeChat Official Account articles and generate theme configuration. Use when the user provides a WeChat article URL and wants to derive a markdown-wechat-converter theme, inject it into markdown-to-wechat.html, or preview the generated result.

Use when saving a Weixin Official Account article from mp.weixin.qq.com into an Obsidian vault as Markdown with locally downloaded images and vault-style asset links.

Export Yuque knowledge bases, Yuque documents, or .lakebook files into local Markdown folders for Obsidian. Use when users want to export Yuque, convert lakebook to Markdown, migrate a Yuque knowledge base to Obsidian, batch-convert multiple .lakebook files, or fix Yuque export issues such as missing images, cropped image mismatches, broken internal links, wrong folder hierarchy, and Markdown table rendering problems.