kennyzheng-builds

Convert Markdown articles to WeChat Official Account (微信公众号) formatted HTML and publish to draft box via API. Full pipeline from markdown to published draft with image support. Use when the user mentions "发公众号", "发布到微信", "微信公众号发布", "publish to wechat", "wechat publish", "公众号排版并发布", or wants to convert markdown to WeChat HTML and publish via API. Also handles format-only (no publish) when user just wants WeChat HTML preview with "公众号排版", "微信排版", "wechat format".

Keep OpenClaw Gateway running 24/7 without manual intervention. Auto-recovery system that restarts crashed gateway, monitors health every 60s, prevents Mac from sleeping, and handles log rotation. ALWAYS use this skill when user mentions: OpenClaw downtime, gateway crashes, service reliability, Mac Mini server setup, 24/7 operation, automatic restart, or keeping services alive. Use proactively if you detect the user has OpenClaw installed and might benefit from stability improvements.

Agent-to-Agent collaboration via shareable links. Use this skill when the user expresses intent to transfer context, problems, knowledge, or work results to another AI agent for collaboration — regardless of language. Core intent patterns: (1) seeking help from another agent, (2) sharing knowledge/context with another agent, (3) handing off work to another agent, or (4) receiving/processing links from agentslink.link. Trigger when the user mentions another agent (e.g., "another agent", "someone's agent", "[role]'s agent", "frontend agent") combined with transfer actions (sending, giving, sharing, packaging, passing, handing off). Example phrases in English: "pack this for another agent", "share this with their agent", "get help from someone's agent". Example phrases in Chinese: "帮我打包给另一个agent", "让xxx的agent了解", "找人帮忙看看", "给前端agent". Also trigger when user pastes agentslink.link URLs. This skill works in any language — if the user is clearly packaging context for inter-agent collaboration, this skill applies.

Seek and analyze video content using Memories.ai's Large Visual Memory Model. Use when the user wants to (1) find or search for videos on TikTok, YouTube, or Instagram by topic, hashtag, or creator, (2) analyze, summarize, or ask questions about a video from a URL, (3) research social media content trends, influencers, or viral patterns, (4) summarize meetings, lectures, or webinars from recordings and extract notes or action items, (5) build a searchable knowledge base from video content and text memories, (6) analyze or describe images with AI vision, (7) store and retrieve text knowledge with semantic search, or (8) import and index social media content into a personal video library.

Transform transcripts, conversations, or raw notes into shareable content — either multiple social posts OR a single long-form article. Learns style from user's published content. Handles transcripts of any length (tested up to 5500+ lines / 90+ min). Use when user: (1) provides transcript/conversation/recording and wants social content, (2) asks to "turn recording/notes into posts", (3) mentions generating content from recordings/notes, (4) needs multiple publishable pieces from single source, (5) asks to write an article from a transcript — e.g. "写成文章/长文", "整理成一篇文章", "write up this session", "turn this talk into a blog post", "深度文章". Social mode outputs 2-15 ready-to-publish pieces. Article mode outputs a structured long-form piece (3000-10000+ chars) written section-by-section.

Dual-purpose vibe coding skill. Mode 1 (Retrospect): Analyzes any completed vibe coding project — codebase, git history, session logs — and generates a human-readable HTML experience article plus a structured knowledge document for agents. Mode 2 (Learn): Takes someone else's project knowledge document, configures CLAUDE.md with teaching rules and project references, so the user's agent can guide them through building a similar project step by step. Use when user says: "复盘", "retrospect", "生成经验贴", "总结开发经验", "打包我的开发经验", "review my project", "write up how I built this", "我想学做这个项目", "帮我学习这个项目", "参考这个项目", "coach me through building this", "learn from this project".

每日 Issue 审阅日报。拉取最近 open issue，检查评论、关联 PR/commit、标签等事件，按处理状态分组，以中文表格呈现。用户说"看 issue"、"issue 日报"、"未回复 issue"时触发。

End-to-end workflow to add Claude OAuth support to Cherry Studio. Applies code changes, reviews, pushes to private GitHub repo, and builds macOS app. Run on a fresh upstream checkout or after git pull.

Extract incremental meeting minutes from transcripts or Feishu documents. Focuses ONLY on discussion-generated information: decisions reached through debate, unresolved disagreements, and action items with assignees. Use this skill whenever the user provides a meeting transcript (text, Feishu doc URL, or audio transcription) and wants meeting minutes, action items, or a summary of what was decided. Also trigger when user says "会议纪要", "会议总结", "整理一下会议", "meeting notes", "meeting summary", "what was decided", or similar.

Scrape any X/Twitter account's posts and generate an interactive analytics dashboard HTML. Use when: (1) user provides an X/Twitter profile URL or username and wants data scraped, (2) user asks to analyze someone's Twitter/X account growth or content, (3) user wants a social media analytics dashboard or visualization, (4) user mentions 'X analytics', 'Twitter analysis', 'scrape tweets', or 'account analysis'. Accepts profile URLs (x.com/username, twitter.com/username), post URLs, or @handles. Outputs a self-contained dark-themed HTML dashboard with Chart.js growth charts, top posts ranking, content analysis, audience voice analysis from real comments, and AI-generated tweet drafts.

Configure OpenClaw model API providers with minimal user input. Use when user wants to add, switch, or connect LLM APIs (e.g., "接入百炼", "配置DashScope", "用DeepSeek", "换个模型"). Auto-detects provider settings from known providers database — user only needs to provide API key and optionally which plan/model they want.