SHILIU · 创作资源
多平台批量内容抓取 · 创作技能
feedgrab 命令行批量抓取工具的使用说明：覆盖X、小红书、Reddit、微信、YouTube、飞书、知乎的书签、时间线、关键词搜索与专辑抓取，含环境变量、登录与断点续传。需自行安装工具并配置账号。
这个技能解决什么问题
feedgrab 命令行批量抓取工具的使用说明：覆盖X、小红书、Reddit、微信、YouTube、飞书、知乎的书签、时间线、关键词搜索与专辑抓取，含环境变量、登录与断点续传。需自行安装工具并配置账号。
适用任务
用 feedgrab 命令批量抓取指定平台的内容，用于选题调研、竞品账号追踪或素材收集。
需要的输入
已安装的 feedgrab 命令行工具；目标平台URL或关键词；对应平台的登录会话或API密钥（如 YOUTUBE_API_KEY、FEISHU_APP_ID/SECRET）；可选时间范围与数量限制。
交付内容
各平台输出目录下的Markdown表格与CSV（如 output/X/search/、output/Reddit/、output/XHS/search/、output/Zhihu/search/），以及可选的单篇.md文件与媒体下载。
任务示例
我已安装 feedgrab 并完成 twitter 登录，请用 x-so 搜索关键词“AI Agent”，取最近7天、点赞100以上，输出表格和CSV。
输出示例
示意结构：命令执行说明 → 输出路径 output/X/search/ 下的Markdown表格（列：作者/内容摘要/点赞/链接）与CSV文件 → 抓取条数与跳过重复项说明。实际结果取决于账号权限与平台返回，此处不含真实抓取数据。
来源与使用
作者 iBigQiang；许可 MIT。已核对来源与包结构，未据此宣称完成运行实测。
机器可读详情 · 读取 SKILL.md · 下载完整技能包
技能说明全文
--- name: feedgrab-batch description: Batch content grabber — bulk fetch bookmarks, user tweets, search results, author notes, wiki pages, subreddit posts, and more from X/Twitter, Xiaohongshu, Reddit, WeChat, YouTube, Feishu, Zhihu. Use when user wants to batch/bulk fetch, search keywords, or grab all posts from an account. --- # feedgrab-batch — Batch Content Grabber > Bulk fetch content from any platform — bookmarks, user timelines, keyword search, album pages, wiki trees. ## Trigger Activate when user mentions batch/bulk operations: - `/feedgrab-batch <command> <args>` - "Grab all my bookmarks" - "批量抓取这个用户的推文" - "搜索关键词 AI Agent" - "抓取这个公众号的所有文章" - "下载这个飞书知识库" ## Prerequisites 1. feedgrab installed (`which feedgrab`) 2. Platform-specific cookies/API keys configured (see each command below) If not ready, suggest `/feedgrab-setup`. ## Command Reference ### X/Twitter #### Bookmarks ```bash # Requires: X_BOOKMARKS_ENABLED=true + feedgrab login twitter (primary session only) feedgrab https://x.com/i/history # All bookmarks (current X URL) feedgrab https://x.com/i/history/bookmarks/2015311287715340624 # Specific folder feedgrab https://x.com/i/bookmarks # Legacy URL, still accepted ``` #### User Tweets ```bash # Requires: X_USER_TWEETS_ENABLED=true + feedgrab login twitter feedgrab https://x.com/username # All tweets X_USER_TWEETS_SINCE=2026-01-01 feedgrab https://x.com/username # Since date # Auto browser search supplement when >800 tweets ``` #### List Tweets ```bash # Requires: X_LIST_TWEETS_ENABLED=true + feedgrab login twitter feedgrab https://x.com/i/lists/LIST_ID # Last 1 day X_LIST_TWEETS_DAYS=7 feedgrab https://x.com/i/lists/LIST_ID # Last 7 days X_LIST_TWEETS_SUMMARY=true feedgrab https://x.com/i/lists/LIST_ID # + summary table ``` #### Keyword Search (x-so) ```bash feedgrab x-so "AI Agent" # Search tweets feedgrab x-so "AI Agent" --days 7 # Last 7 days feedgrab x-so "AI Agent" --min-faves 100 # Min 100 likes feedgrab x-so WorkBuddy --lang zh+zxx --sort all # Chinese tweets + Article cards, merged Latest/Top feedgrab x-so "claude,cursor,copilot" --merge # Multi-keyword merged table feedgrab x-so "AI Agent" --raw # Raw query syntax ``` Output: Markdown table (sorted by views) + CSV at `output/X/search/` --- ### Reddit #### Subreddit Batch ```bash feedgrab reddit-sub MachineLearning --sort hot --limit 25 feedgrab reddit-sub ChatGPT --sort top --limit 50 ``` #### Keyword Search (reddit-so) ```bash feedgrab reddit-so "codex" --sort comments --time all --limit 10 feedgrab reddit-so "AI Agent" --subreddit ChatGPT --sort top --time week REDDIT_SEARCH_SAVE_POSTS=true feedgrab reddit-so "open source" --limit 5 ``` Output: Markdown summary at `output/Reddit/`. Use `REDDIT_REPLY_MODE=tree|all` when deep-fetching posts. --- ### Xiaohongshu (小红书) #### Author Notes ```bash # Requires: XHS_USER_NOTES_ENABLED=true + feedgrab login xhs feedgrab https://www.xiaohongshu.com/user/profile/USER_ID XHS_USER_NOTES_SINCE=2026-01-01 feedgrab https://www.xiaohongshu.com/user/profile/USER_ID ``` #### Search Results ```bash # Requires: XHS_SEARCH_ENABLED=true + feedgrab login xhs feedgrab "https://www.xiaohongshu.com/search_result?keyword=开学第一课&source=web_explore_feed" ``` #### Keyword Search (xhs-so) ```bash feedgrab xhs-so "AI Agent" # Search notes feedgrab xhs-so "AI Agent" --sort popular # By popularity feedgrab xhs-so "AI Agent" --type video # Videos only feedgrab xhs-so "AI Agent" --save # Also save individual .md feedgrab xhs-so "claude,cursor" --merge # Multi-keyword merged ``` Output: Markdown table (sorted by likes) + CSV at `output/XHS/search/` --- ### WeChat (微信公众号) #### By Account (mpweixin-id) ```bash # Requires: feedgrab login wechat (MP backend session, ~4 day validity) feedgrab mpweixin-id "公众号名称" MPWEIXIN_ID_SINCE=2026-01-01 feedgrab mpweixin-id "公众号名称" # Since date ``` #### Album / Collection (mpweixin-zhuanji) ```bash feedgrab mpweixin-zhuanji "ALBUM_URL" # No login needed MPWEIXIN_ZHUANJI_SINCE=2026-01-01 feedgrab mpweixin-zhuanji "ALBUM_URL" ``` #### Sogou Search (mpweixin-so) ```bash feedgrab mpweixin-so "关键词" # Search via Sogou feedgrab mpweixin-so "关键词" --limit 30 # Max 30 results ``` --- ### YouTube #### Search (ytb-so) ```bash # Requires: YOUTUBE_API_KEY feedgrab ytb-so "machine learning" # Search videos feedgrab ytb-so "AI" --order viewCount # Sort by views feedgrab ytb-so "AI" --channel @3blue1brown # Channel-specific feedgrab ytb-so "AI" --days 30 # Last 30 days ``` #### Download ```bash feedgrab ytb-dlv "VIDEO_URL" # Download video (MP4) feedgrab ytb-dla "VIDEO_URL" # Download audio (MP3) feedgrab ytb-dlz "VIDEO_URL" # Download subtitles (SRT) ``` --- ### Feishu/Lark (飞书) #### Wiki Batch ```bash # Requires: FEISHU_APP_ID + FEISHU_APP_SECRET feedgrab feishu-wiki "WIKI_SPACE_URL" ``` --- ### Zhihu (知乎) #### Keyword Search (zhihu-so) ```bash # Requires: feedgrab login zhihu (for full content) feedgrab zhihu-so "AI Agent" # Search Q&A + articles feedgrab zhihu-so "AI Agent" --sort hot # By popularity feedgrab zhihu-so "AI Agent" --limit 20 # Max 20 results feedgrab zhihu-so "AI Agent" --days 30 # Last 30 days feedgrab zhihu-so "claude,cursor" --merge # Multi-keyword merged ``` Output: Markdown table (sorted by upvotes) + CSV at `output/Zhihu/search/` --- ## Environment Variables Quick Reference | Platform | Required Variables | How to Get | |----------|-------------------|------------| | Twitter (basic) | Cookie via `feedgrab login twitter` | Login in browser | | Twitter (paid API) | `TWITTERAPI_IO_KEY` | twitterapi.io | | Reddit | Optional Cookie via `feedgrab login reddit`; `REDDIT_SEARCH_*`, `REDDIT_REPLY_MODE` | Login only if Cloudflare/private content requires it | | Xiaohongshu | Cookie via `feedgrab login xhs` | Login in browser | | WeChat batch | Cookie via `feedgrab login wechat` | Login MP backend | | YouTube | `YOUTUBE_API_KEY` | Google Cloud Console | | GitHub | `GITHUB_TOKEN` (optional) | GitHub Settings | | Feishu | `FEISHU_APP_ID` + `FEISHU_APP_SECRET` | Feishu Open Platform | | Zhihu | Cookie via `feedgrab login zhihu` | Login in browser | ## Resume After Interruption All batch commands support **checkpoint resume** — if interrupted, just re-run the same command. Already fetched items are skipped via dedup index. ## Tips - Use `feedgrab doctor <platform>` to diagnose configuration issues - Media download: set `X_DOWNLOAD_MEDIA=true` / `XHS_DOWNLOAD_MEDIA=true` / `MPWEIXIN_DOWNLOAD_MEDIA=true` - Multi-account cookie rotation for Twitter: put extra cookies in `sessions/x_2.json`, `x_3.json`
技能是任务材料，不会自动授予账户权限、付费授权或其他系统权限。请让 Agent 按说明核对依赖，再在授权范围内使用。