xiaobei

Publish videos to Bilibili (B站) via Open Platform API (OAuth2). Supports chunked video upload, cover image, tags, and partition selection. Requires BILIBILI_APP_ID and BILIBILI_APP_SECRET environment variables.

内容校准预测循环——打分 → 盲预测 → T+3d复盘 → 进化 rubric。按平台独立迭代，每个平台拥有自己的 rubric、校准池、预测日志、受众画像。本技能负责打分（blind sub-agent + score-only.sh + 阈值门 + 流程 1A）与校准闭环；发布记录与数据采集由 published-track 统一管理。

口播视频去口误。自动识别并删除静音、语气词、卡顿词、重复句、残句等，输出干净视频+字幕+剪映草稿。触发词：去口误、剪口播、de-mouth、去除口误

Generate custom WeChat CSS themes from natural language descriptions, WeChat article URLs, or recent articles from a WeChat Official Account. Produces a valid CSS file conforming to wenyan and ready for wx-mp-publisher.

一站式短视频制作。支持文章链接、追爆报告、文字主题等多种输入，用 gen.py 直连火山 Seedance / 阿里云百炼 Wan2.7-HappyHorse 端点生成视频素材（声画同出），FFmpeg 组装成片。

通过浏览器自动化发布视频到微信视频号。处理 wujie shadow DOM，支持视频上传、标题描述填写、即时发布。

通过浏览器自动化在微博发布图文内容。微博 API 对个人开发者不友好，浏览器方案更实用。

Publish content (text + images/video/link) to WeChat Work (企业微信) customer moments via a single bash script. Supports local mode (IP whitelisted) and relay mode via tx-relay proxy.

Upload videos to YouTube via YouTube Data API v3. Supports video upload (resumable), title, description, tags, and visibility settings. Requires Google OAuth2 credentials.

通过浏览器自动化在知乎发布文章或回答。知乎无可用公开 API，需通过浏览器操作完成发布。

为商业拓展/投资人关系建立场景创建精美的 HTML 演示文稿——路演 PPT、合作提案、客户 Demo、产品介绍。零依赖单文件输出，可直接通过邮件/微信发送或浏览器演示。

智能搜索路由器。根据用户意图选择最佳搜索源，构造 URL 并导航，获取内容。 Replaces the built-in web_search tool for targeted, platform-specific searches.

Invite a qualified customer into the experience group when they want to understand the product form further after seeing demo materials. The invite is sent as an awada control message, and the customer status is updated to exp_invited to prevent duplicate invitations.

Publish content to Douyin (抖音) via open platform H5 Schema. Generates a schema URL to open Douyin app's publish page. Supports video, images, album, hashtags, privacy, note mode, and forward-to-daily. Requires Douyin open platform web app credentials.

Publish posts, videos, and reels to Facebook via Meta Graph API v23.0. Supports feed posts, video posts, photo posts, reels, and stories. Requires Meta OAuth2 page access token.

自动从本地视频中提取高光片段。通过 ASR 转录 + 文本分析识别高光时刻，剪辑输出多段短视频。

Publish posts and reels to Instagram via Meta Graph API. Supports single images, carousels (up to 10), and reels. Uses content container pattern. Requires Meta OAuth2 token with instagram_basic and instagram_content_publish permissions.

发布文章到掘金平台。使用浏览器自动化完成发布流程，包括在线编辑器输入内容、选择分类、添加标签、上传封面图、发布。当用户要求发布内容到掘金时触发。

Create pins on Pinterest via Pinterest API v5. Supports image pins and video pins with board selection. Requires Pinterest OAuth2 token.

Publish posts to Threads via Meta Graph API. Supports text, image, video, and carousel posts. Requires Meta OAuth2 token with threads_basic and threads_content_publish permissions.

Upload videos to TikTok via Content Posting API. Supports direct upload and URL-based publish, hashtags, and privacy settings. Requires TikTok OAuth2 credentials.

Publish Markdown articles to 今日头条 via docx document import. Converts Markdown (with local images embedded) to docx, then guides through Toutiao's "文档导入" upload flow.

Compose and publish a post (text, image, or video) to Twitter/X using the browser. Supports single and thread posts.

Render and publish Markdown articles to WeChat Official Account (公众号) draft box. Supports local mode (IP whitelisted) and server relay mode via Wenyan Server. Supports multi-account, image-only posts (小绿书), and proxy.

邮件操作技能。处理策略性单封邮件：回复询盘、跟进邮件、会议确认、合作提案草稿，以及发送后的状态核实。与 cold-outreach 互补——cold-outreach 负责批量自动化冷邮件，email-ops 负责有具体目标对象的关键邮件。

Search and download copyright-free images and video clips from Pexels API. Supports both photo search (--type image) and video search (--type video, default).

Search and download copyright-free images and video clips from Pixabay API. Alternative to pexels-footage when Pexels has no suitable results. Supports both photo search (--type image) and video search (--type video, default).

从文稿内容生成专业 PPTX 演示文稿，支持用户提供的模板/参考图风格提取、AI 配图（siliconflow-img-gen）和素材图库（pexels/pixabay）。纯 Python 生成，无需 Google/ChatGPT API。

Discover the RSS/Atom feed URL for a website, then run the fetch-rss.mjs script to retrieve and parse articles from the feed.

Generate or edit images via SiliconFlow Images API. Text-to-image uses Qwen/Qwen-Image; image-edit uses Qwen/Qwen-Image-Edit-2509.

网络表单填报技能 — 从信息搜集到浏览器填报的完整工作流，确保框架受控组件正确写入、页面切换前暂存、最终提交由用户确认。

小红书社交互动技能。发表评论、回复评论、点赞。当用户要求评论、回复或点赞小红书帖子时触发。

Maintain Main Agent reminder.json for onboarding, work channel recommendations, HRBP enablement, Media Operator bootstrap, and Gateway restart followups.

软件著作权登记全流程：从代码仓/目录生成程序鉴别材料（源程序文档）、 软件操作手册、申请填报信息 Markdown，并可辅助在线填报。 当用户需要申请软件著作权、生成软著材料时触发。

维护 business-developer 的 SQLite 追踪数据库，记录已探索的创作者（模式一）和已互动的帖子（模式二），避免重复追踪和重复互动。

通过自媒体平台搜索内容，在评论区以留言/回复/私信等方式拓展潜在客户或进行品牌宣传。用于 HEARTBEAT 定时任务。

维护 business-developer 的 SQLite 情报采集数据库，记录已采集的信息内容，避免重复采集，支持按日查询已采集情报。

定时监控特定信源（自媒体账号/网页），按预设标准提取商业情报，生成简报或报告。用于 cron 定时任务。

通过自媒体平台按搜集策略探索潜在客户——策略 A 分析帖子发布者画像，策略 B 从评论区挖掘潜客。用于 HEARTBEAT 定时任务。

根据风格描述从内置设计系统库中选取最匹配的设计系统，提供 DESIGN.md 规范供后续设计工作使用。