Invoice Analyzer
Analyze invoices in财务管理 directory and extract key information. Update financial list with invoice status. Use when user mentions invoice recognition, 开票状态, or checking which items have been invoiced.
Paper Html Onepage
>
Assignment Word
Edit and fill Word (.docx) documents — inspect structure, replace text, batch fill placeholders, insert images, and fill tables. Use when user mentions "assignment-word", "docx_editor", "填Word", "填docx", "Word填写", "replace docx", "fill docx", or needs to programmatically modify a Word document.
Git Push
Push git changes to GitHub. Use whenever user wants to upload/commit/push code changes to GitHub repository.
Claude2anti
Synchronize Claude Code user skills into Antigravity. Auto-discovers the Claude environment and syncs from the Claude skills folder into %USERPROFILE%\.gemini\antigravity\skills. Supports full sync and single-skill sync.
Unilab Train
>-
Ai Gen
AI 生图/生视频。通过 paratera API 生成图片和视频。Use when user mentions: ai-gen, 生图, 生视频, 文生图, 文生视频, 生成图片, 生成视频, AI绘图, AI视频, AI生图, AI生视频, image generation, video generation, 批量生图, 批量生视频.
Any2md
Universal document to Markdown converter with automatic format routing. Auto-detects DOCX math and routes to word2md; uses markitdown for PPTX/XLSX/HTML/EPUB/etc; supports PyMuPDF for high-fidelity PDF. Fixes markdownify math-escaping bugs.
Html2pdf
Convert HTML, especially slide-style HTML presentations, into paginated PDF using Chrome or Edge headless printing. Use when the user mentions `html2pdf`, wants to export HTML to PDF, or needs each slide/page in a presentation HTML to become a separate PDF page instead of one long page.
Unilab Rollout Trace
>-
Claude2codex
Synchronize Claude Code user skills into Codex. Auto-discovers the Claude environment and syncs from the Claude skills folder into %USERPROFILE%\.codex\skills. Supports full sync and single-skill sync. Preserves Codex built-in .system skills and accepts both skill.md and SKILL.md.
Assignment
大学作业管理系统。读取作业文档转为 Markdown，用 `/assignment complete` 自动完成解答，用 `/assignment done` 编译 PDF，并支持用 `/assignment figure` 插入原题图。
Paper Repro
>
Plot Train Z1
Generate and sync Z1 12DOF training learning curve plots from TensorBoard data on the RTX server, then compile a single-page A4 PDF report with plots + data analysis. Use when user mentions "plot training", "learning curve", "generate plots", "update plots", "学习曲线", "生成图表", "训练曲线", "训练报告", "生成报告".
Web Search Fallback
>
Sys Health
>-
Word2pdf
Convert Word (.doc / .docx) to PDF. Auto-detects binary .doc (OLE2) vs .docx (Office Open XML) via magic bytes, then converts using Microsoft Word COM or LibreOffice headless. Use when user mentions "word2pdf", "Word转PDF", "Word to PDF", "doc转PDF", "docx转PDF", or needs to convert a Word document to PDF.
Github Trending
>
Gpu Train
>-
Merge
Merge multiple videos into a single grid video (3x1 vertical, 2x2, 3x2 layouts). Supports custom labels per cell. Use when user mentions "merge videos", "合并视频", "拼视频", "视频对比", "video grid", "视频拼接", "compare videos", "side by side video".
Start Codex
>-
Gpu Connect
>-
Rtx Training Capture
>-
Md2docx
>-
Reflection2xhs
Convert a local reflection markdown file, especially `reflection.md` or `*_reflection.MD`, into a polished single-page study-note-style A4 poster workflow. Use when the user wants to turn personal paper reflections, study notes, or structured markdown summaries into a shareable visual poster. This skill is Markdown-first and defaults to using the built-in image generation capability after local p…
Codex2claude
Synchronize Codex user skills into Claude Code. Auto-discovers the Claude environment and syncs from %USERPROFILE%\.codex\skills into the Claude skills folder. Supports full sync and single-skill sync. Excludes Codex built-in .system and accepts both skill.md and SKILL.md.
Latex Beamer
>-
Resume Builder
从用户上传的简历资料(PDF/Word/Markdown/已有 YAML/口述)生成排版好的单栏中文简历 PDF。全程不依赖 LaTeX/MiKTeX/Overleaf——用 HTML 模板 + 系统 Chrome/Edge 无头打印出 PDF。当用户提到「简历」「resume」「cv」「生成简历 PDF」「套用我的简历模板」时使用。
Word2md
Convert Word (.docx) files with Office Math formulas to clean Markdown with LaTeX. Handles OMML to LaTeX conversion, Greek letters, subscripts, superscripts, fractions, and display/inline math.
Project Managing
>-
Cheatsheat
Convert Markdown notes to compact A4 CHEATSHEET format with 3-column layout, 小六 font size (10.5px), and narrow margins (10mm). Images occupy 95% of column width. Use this skill when user mentions cheatsheet, 小抄, 速查表, compact print, or 复习资料.
Pdf Reader
Use PyMuPDF (fitz) to read and extract text from PDF files, instead of the native Read tool. Use this skill when user mentions "read PDF", "读取PDF", "PDF内容", "提取PDF", "PDF text", "open PDF", "打开PDF", or when you need to read any .pdf file. Always prefer this over the native Read tool for PDF files.
Training Log Update
Append or create the daily training-log entry (one YYYY-MM-DD.md per day) in the established house style — a one-line Headline plus explicit Done / Issues-and-fixes / Pending / Key-paths sections, with NO media copied in (reference real paths). Use whenever the user says "update the training log", "记录一下今天的 training log", "写 training log", "log today's work", or wants to record what was done vs wh…
Lab Report
实验报告改写工具（v5.0）。从他人的实验报告生成自己的版本，MD模板→填写→图表→PDF。命令行交互，图片/数据/主观题通过Prompt询问路径。
Img Reader
读取和分析本地图片文件，零 MCP 额度。3级自动降级：Read+PIL增强 → GLM-4V-Flash(免费API) → PaddleOCR(离线)。Use when user mentions: 看图, 读图, 分析图片, 识别图片, img, image, 看一下图片, 图片内容, 截图, screenshot, 照片, photo, 读取图片, image analysis, 明细表, BOM, OCR.
Html Slides
A Claude skill from phanghonghao/THU-Awesome-Skills.
Jianying Editor
剪映 (JianYing) AI自动化剪辑的高级封装 API (JyWrapper)。提供开箱即用的 Python 接口，支持录屏、素材导入、字幕生成、Web 动效合成及项目导出。
Embodied Ai Agent
调研和整理 Bilibili 与 GitHub 上的中文具身智能相关内容，输出按课题路由的中文 Wiki 草稿，并在需要时继续推送到飞书。用户提到 Embodied-Ai-agent、B 站具身智能视频、GitHub 仓库调研、Embodied AI 视频整理、VLA/WM/GR00T/ALOHA 课题归档、视频转 Wiki、仓库转 Wiki、飞书知识库整理时使用。若用户没有提供 VideoUrl 或 RepoUrl，但希望 agent 自己找内容，也使用本 skill：先向用户索取 2-5 个中文关键词，再检索候选内容并继续调研。
Notes
Transform Markdown notes with renamed images, auto-add heading markers, and optionally add dual-column layout for compact PDF output. Use this skill whenever the user mentions notes, markdown formatting, image renaming, adding headings, or organizing note files. Always use this skill when working with Markdown notes that need formatting or image processing.
Pdf2word
Convert PDF files to Word (.docx) while preserving layout, tables, images, and formatting. Use when user mentions "pdf2word", "PDF转Word", "PDF to Word", "PDF转docx", or needs to convert a PDF to an editable Word document.
Poster
HTML/CSS A4 海报生成。从沙龙/活动 Markdown 文件生成科技风 A4 海报 PDF/PNG。Use when user mentions: poster, 海报, 海报生成, make poster, generate poster, 活动海报, 沙龙海报.
Html2tex
>-
Which Key
识别当前 Claude Code 会话使用的是哪个 API Key。Use when user mentions "which key", "哪个key", "当前key", "what key", "用的哪个key", "key是哪个", "当前密钥", "current key".
Markitdown
Convert various document formats to Markdown using Microsoft's markitdown library. Supports PDF, DOCX, PPTX, XLSX, HTML, EPUB, images, IPYNB, ZIP and more. Use when user mentions markitdown, convert to markdown, PPTX to md, XLSX to md, EPUB to md, or needs to read formats not covered by /pdf-reader or /word2md.
Start Claude
Manage and launch Claude Code instances with different API keys. Use --continue to switch key in-place while preserving conversation history. Supports session recovery from other keys. Use when user mentions "start claude", "launch", "switch key", "换key", "切换key", "continue", "保留会话", "恢复会话", "recover session".
Autodl 5090
>-
Venv
Python 版本切换 + 虚拟环境管理。快速创建/激活指定 Python 版本的 .venv。Use when user mentions: /venv, venv, virtualenv, python env, python version, switch python, 切换python, python版本.
Certificate
从沙龙海报自动生成电子证书 PNG。Use when user mentions: certificate, 证书, 电子证书, 生成证书, generate certificate, 名人堂证书, 沙龙证书.
Md2tex
>-
Word2html
Convert Word (.docx) files to standalone HTML with tables, images, and math formulas. Handles merged cells (colspan/rowspan), embedded images, bold/italic/underline, and OMML math via MathJax.