by chubbyguan
把中文全渠道内容(抖音 / B站 / 小红书 / 公众号 / X / 播客)采集进个人知识库的 13 个 AI Skill:图文存图、视频转文字稿、字幕优先免 GPU,附带知识库 MCP server。 | Ingest Chinese content into your personal knowledge base — image/video routing, subtitle-first transcription, and a KB MCP server.
# Add to your Claude Code skills
git clone https://github.com/chubbyguan/chubbyskillsLast scanned: 6/15/2026
{
"issues": [],
"status": "PASSED",
"scannedAt": "2026-06-15T10:24:00.672Z",
"npmAuditRan": true,
"pipAuditRan": false,
"promptInjectionRan": true
}See how chubbyskills compares with popular alternatives.
chubbyskills is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by chubbyguan. 把中文全渠道内容(抖音 / B站 / 小红书 / 公众号 / X / 播客)采集进个人知识库的 13 个 AI Skill:图文存图、视频转文字稿、字幕优先免 GPU,附带知识库 MCP server。 | Ingest Chinese content into your personal knowledge base — image/video routing, subtitle-first transcription, and a KB MCP server. It has 671 GitHub stars.
Yes. chubbyskills passed SkillsLLM's automated security scan — a dependency vulnerability audit plus prompt-injection heuristics — with no high-severity issues. You can read the full report in the Security Report section on this page.
Clone the repository with "git clone https://github.com/chubbyguan/chubbyskills" and add it to your Claude Code skills directory (see the Installation section above).
chubbyskills is primarily written in Python. It is open-source under chubbyguan on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh chubbyskills against similar tools.
No comments yet. Be the first to share your thoughts!
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM is an informational directory and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
中文 · English
Agent Skills 生态已有 140 万+ 技能包,但绝大多数是泛化开发技能;垂直领域(中文内容采集 → 知识库)至今没有事实标准。Chubby Skills 要占据的,就是这个位置。
| 维度 | Chubby Skills | feedgrab | RSSHub | 商业工具(Readwise / ima / NotebookLM) |
|---|---|---|---|---|
| 中文平台全渠道采集 | ✅ 10 平台 | ⚠️ 7 平台 | ⚠️ 订阅源(非正文采集) | ⚠️ 覆盖差 / 收费 |
| 视频 / 播客转录(字幕优先免 GPU) | ✅ | ❌ | ❌ | ⚠️ 部分 |
| 知识库 + 语义检索 + MCP 闭环 | ✅ 完整 | ❌ 只抓不存 | ❌ | ⚠️ 封闭生态 |
| 本地运行 / 隐私 | ✅ 完全本地 | ✅ 本地 | ✅ | ❌ 云端 |
| 可被 Agent 编排(开放标准) | ✅ | ✅ | ❌ | ❌ |
| 免费 / 零 API 费用 | ✅ 零依赖档位可用 | ✅ | ✅ | ❌ 订阅制 |
采集工具有很多,知识库工具也有很多。但「中文全渠道采集 → 统一格式 → 知识库 → Agent 调用」的完整闭环、且完全本地可迁移的,只有 Chubby Skills。
完整对比与生态定位图见 docs/comparison.md。
Chubby Skills 是一套面向个人知识库和 AI Agent 的内容采集、整理、检索工具。
你可以把它理解成三层:
一句话:把你每天刷到、听到、读到的好内容,变成可复用、可检索、可被 Agent 调用的个人知识资产。
当前版本:0.11.0
| 模块 | 能力 | 入口 |
|---|---|---|
| 平台采集 | 视频、图文、公众号、播客、X、小红书等内容转 Markdown | tools/chubby_ingest.py / 各 skill 脚本 |
| 管线编排 | 队列、状态、重试、日报、schema v1 元数据 | tools/chubby.py |
| 内容加工 | 摘要、要点、标签、价值判断 | content-enrich |
| 知识库 | vault 模板、SQLite 索引、全文搜索、语义检索、最近笔记、统计 | tools/vault_index.py |
| 知识自动化 | 自动归档、知识卡片生成 | tools/vault_curator.py |
| MCP | Agent 搜索、读取、重建索引、查看统计 | knowledge-base-management/scripts/mcp_server.py |
| 质量保障 | 平台健康度、smoke matrix、golden outputs、schema 校验 | tools/platform_health.py / tools/platform_smoke.py / tools/golden_outputs.py |
| 贡献者适配 | 新平台 definition / template / skill scaffold | tools/platform_adapter.py |
git clone https://github.com/chubbyguan/chubbyskills.git
cd chubbyskills
python3 tools/chubby.py quickstart
quickstart 是离线首跑验收,不会抓真实平台内容。它会检查:
发布级验收:
python3 tools/chubby.py --version
python3 tools/platform_smoke.py --mode all --check
python3 tools/golden_outputs.py examples/outputs
python3 tools/mcp_workflow_demo.py
更多发布检查见 docs/release.md。
bash setup.sh # 默认 light:轻量能力
bash setup.sh video # 视频转录重依赖
bash setup.sh podcast # 播客转录
bash setup.sh wechat # 公众号/PDF 处理
bash setup.sh all # 全部依赖
bash setup.sh doctor # 只做环境体检
轻量模式可直接使用:X 图文、小红书图文、公众号基础处理、行业情报雷达、知识库健康检查、content-enrich。视频和播客转录才需要 ffmpeg、yt-dlp、funasr、torch、faster-whisper 等重依赖。
pip install -r requirements.txt
pip install -r podcast-transcribe/requirements.txt
每个 skill 都是独立目录,也可以只安装你需要的那个。
在 Claude Code、Codex、OpenClaw、Hermes 等支持 Skill 的 Agent 里,可以直接说:
帮我安装这个 skill:https://github.com/chubbyguan/chubbyskills/tree/main/<skill-name>
Gitee 镜像:https://gitee.com/chubbyguan/chubbyskills
| 平台 / 内容 | Skill | 能力 | 默认依赖 | 失败 fallback |
|---|---|---|---|---|
| B站 | bilibili-transcribe |
字幕优先,视频转录,批量 URL | yt-dlp,无字幕需 ffmpeg + funasr |
无字幕时走音频转录 |
| YouTube | youtube-transcribe |
字幕优先,转录,翻译,中英对照,批量 URL | yt-dlp,翻译需 DEEPSEEK_API_KEY |
无字幕时走音频转录 |
| 抖音 | douyin-transcribe |
视频转文字稿 | ffmpeg + funasr |
本地视频转录 |
| TikTok | tiktok-transcribe |
视频转文字稿 | yt-dlp + ffmpeg + funasr |
本地视频转录 |
| 微博 | weibo-transcribe |
微博视频转文字稿 | yt-dlp + ffmpeg + funasr |
优先移动端链接或本地视频 |
| 知乎 | zhihu-transcribe |
知乎视频转文字稿 | yt-dlp + ffmpeg + funasr |
本地视频转录 |
| 播客 | podcast-transcribe |
小宇宙 / 喜马拉雅 / RSS / 本地音频转录 | ffmpeg + faster-whisper |
本地音频转录 |
| 微信公众号 | wechat-article-ingest |
公众号文章 / PDF 转 Markdown | beautifulsoup4,PDF 增强需 markitdown / pymupdf |
PDF 或保存 HTML |
| 小红书 | xiaohongshu-ingest |
图文存图、视频转录、爆款拆解、衍生选题 | 图文零依赖,建议 XHS_COOKIE |
--fallback-text 手动正文 |
| X / Twitter | x-ingest |
推文正文、图片、视频转录,免 API Key | 图文零依赖,视频需 ffmpeg + funasr |
--fallback-text 手动正文 |
平台状态页:docs/platform-status.md
失败分型和 fallback 指南:docs/platform-fallbacks.md
python3 tools/chubby_ingest.py "https://www.bilibili.com/video/BVxxxx" -o output/
python3 tools/chubby_ingest.py "https://x.com/user/status/123" -o output/
python3 tools/chubby_ingest.py "https://mp.weixin.qq.com/s/xxx" -o output/
自动识别平台失败时,可以显式指定:
python3 tools/chubby_ingest.py "<链接>" --skill youtube -o output/
export DEEPSEEK_API_KEY="..."
python3 tools/chubby_ingest.py "<链接>" -o output/ --enrich
content-enrich 会补摘要、要点、标签、领域和价值判断,并保留原文。
python3 tools/chubby_ingest.py "<链接>" --vault ~/Documents/Obsidian/Inbox
也可以使用完整管线:
python3 tools/chubby.py init
python3 tools/chubby.py ingest "<链接>"
python3 tools/chubby.py status --latest
python3 tools/chubby.py retry --all-failed
tools/chubby.py 会记录:
.chubby/runs.jsonl:每条 source 的状态、错误、输出路径、run_idruns/YYYY-MM-DD.md:每日运行报告run_id、source_hash、captured_at、processed_at、content_type、assets把链接逐行放进 inbox/links.txt:
https://x.com/user/status/123
https://mp.weixin.qq.com/s/xxx
https://www.bilibili.com/video/BVxxxx
运行:
python3 tools/chubby.py run
python3 tools/chubby.py status --failed
python3 tools/chubby.py retry --all-failed
空行和 # 注释会被忽略。
python3 tools/vault_index.py index ~/Documents/ObsidianVault
python3 tools/vault_index.py search "AI Agent"
python3 tools/vault_index.py semantic "内容策略"
python3 tools/vault_index.py recent --limit 10
python3 tools/vault_index.py stats
读取某篇笔记:
python3 tools/vault_index.py read "10_Sources/x/example.md" --vault ~/Documents/ObsidianVault
默认语义检索是零依赖 semantic-lite。如果你想用真实向量:
OpenAI:
export OPENAI_API_KEY="..."
python3 tools/vault_index.py embed ~/Documents/ObsidianVault --provider openai
python3 tools/vault_index.py semantic "内容策略" --provider openai
本地模型:
python3 -m pip install sentence-transformers
python3 tools/vault_index.py embed ~/Documents/ObsidianVault --provider local
python3 tools/vault_index.py semantic "内容策略" --provider local
更多说明见 docs/knowledge-automation.md。
默认 dry-run,不会移动文件:
python3 tools/vault_curator.py archive ~/Documents/ObsidianVault
确认后真正执行:
python3 tools/vault_curator.py archive ~/Documents/ObsidianVault --apply
python3 tools/vault_curator.py card ~/Documents/ObsidianVault "10_Sources/x/example.md" --apply
归档规则:
00_Inbox/**/*.md 是待处理区。summary、archive_status: processed、processed / evergreen 标签的笔记进入 20_Processed/。10_Sources/<platform>/。启动 MCP server:
pip install mcp
VAULT_DIR=~/Documents/ObsidianVault python3 knowledge-base-management/scripts/mcp_server.py
MCP 工具:
search_vaultsemantic_search_vaultread_kb_notelist_recent_notesreindex_vaultvault_index_stats示例工作流:
python3 tools/mcp_workflow_demo.py
这个 demo 会用 fixtures/mcp-vault 完成一次“检索 vault → 读取原文 → 带来源回答”的 Agent 任务。详细配置见 docs/mcp-workflow.md。
这个仓库现在不是只提供脚本,也提供可复现的验收层。
| 检查 | 命令 | 作用 |
|---|---|---|
| 输出协议 | python3 tools/validate_outputs.py examples/outputs --schema-v1 |
确认 Markdown frontmatter 和 schema v1 |
| 平台定义 | python3 tools/platform_health.py --check |
校验 platforms/*.yaml、模板和脚本路径 |
| 状态页新鲜度 | python3 tools/platform_health.py --check-output |
确认 docs/platform-status.md 未过期 |
| 平台 smoke | python3 tools/platform_smoke.py --mode all --check |
验证 offline / fallback / live 分层 |
| Golden outputs | python3 tools/golden_outputs.py examples/outputs |
防止示例输出结构被意外改坏 |
| MCP workflow | python3 tools/mcp_workflow_demo.py |
验证 Agent 能通过 vault 完成任务 |
live smoke 是显式 opt-in,因为真实平台会受 cookie、地区、风控、重依赖和链接有效期影响:
export CHUBBY_SMOKE_X_SOURCE='https://x.com/<user>/status/<id>'
python3 tools/platform_smoke.py --mode live --check
所有采集类 skill 都输出带 frontmatter 的 Markdown,基础字段包括:
| 字段 | 含义 |
|---|---|
title |
标题 |
type |
类型,通常是 note |