Ingest Chinese social and media content into agent knowledge bases with Chubbyskills
Use Chubbyskills when an agent needs installable skills for collecting Chinese social, video, podcast, and article content into Markdown and Obsidian-style knowledge bases.
npx skills add agentskillexchange/skills --skill ingest-chinese-social-and-media-content-into-agent-knowledge-bases-with-chubbyskills
Use Chubbyskills when an operator wants an agent to turn links from Chinese and global content channels into structured local knowledge instead of manually saving screenshots, copying transcripts, and cleaning notes. The workflow is to install only the needed skill, provide a supported source link such as Douyin, Bilibili, Xiaohongshu, WeChat article, X, YouTube, TikTok, Weibo, Zhihu, or a podcast source, let the skill route image or video content, prefer subtitles where available, transcribe audio when needed, and write normalized Markdown with consistent frontmatter for later retrieval.
What this skill actually does
Invoke this instead of using each platform normally when the task is repeatable capture into a personal or team knowledge base: save a video as a transcript, preserve a social post with images, enrich an imported note with summary and tags, run a health check on the vault, or maintain a daily industry-intelligence radar. The scope boundary is local content ingestion and knowledge-base maintenance through explicit agent skills. It is not a general scraping service, not permission to bypass platform terms, not a commercial content harvesting workflow, and not a generic MCP server listing.
Inputs and prerequisites: a supported agent runtime such as Claude Code, Codex, OpenClaw, OpenCode, or Hermes; the selected Chubbyskills skill folder; optional ffmpeg, yt-dlp, SenseVoice, faster-whisper, or DeepSeek API access depending on the selected skill; and an approved local Markdown or Obsidian-style knowledge base.
Setup notes: Clone the repository or install a specific skill folder, run the documented setup path for all skills or the chosen individual skill, install only the dependencies required by that platform workflow, configure any optional API key locally, and test with a single approved source link before routine use.
Source and verification boundary: use https://github.com/chubbyguan/chubbyskills as the canonical reference before running the workflow; keep source links, downloaded artifacts, transcription outputs, enrichment steps, and knowledge-base writes reviewable against the upstream skill instructions.
Framework fit: publish this as a Multi-Framework workflow because upstream explicitly ships Agent Skills compatible with several agent runtimes and also includes a knowledge-base MCP server, while invocation is through installable skill folders rather than one runtime-specific SDK.