agent-browser
Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, test web applications, or extract information from web pages.
commit
Use when preparing or executing git commits in this repo and the commit message must follow the emoji conventional template, especially for auto-commit or "just commit" requests that might bypass it.
frontend-design
Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, artifacts, posters, or applications (examples include websites, landing pages, dashboards, React components, HTML/CSS layouts, or when styling/beautifying any web UI). Generates creative, polished code and UI design that avoids generic AI aesthetics.
skill-creator
Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.
tapestry
Unified content extraction and action planning. Use when user says "tapestry <URL>", "weave <URL>", "help me plan <URL>", "extract and plan <URL>", "make this actionable <URL>", or similar phrases indicating they want to extract content and create an action plan. Automatically detects content type (YouTube video, article, PDF) and processes accordingly.
archlinux-desktop-ops
仅用于 mainliufeng 的 Arch Linux / Hyprland 笔记本。用于管理 Hyprland workspace/window、蓝牙耳机连接与音频路由、触摸板异常恢复，以及调用 dotfiles 中已有桌面脚本；不要在 macOS 上使用。
chrome-access-routing
Route Chrome, browser-access, and web-research tasks across Codex and Hermes on macOS or Linux, including Chrome Extension, Computer Use, Chrome DevTools Protocol, Browser, web-access, agent-browser, and headless/browser MCP tools. Use when the user asks an agent to research, open, inspect, control, screenshot, QA, or debug web pages, especially authenticated dashboards, the user's already-open Chrome tab, or cases where one route fails and another should be tried before giving up.
computer-use-non-disruptive
Use before any Computer Use / @Computer / desktop app GUI operation, especially on macOS with Spaces/workspaces, multiple windows, background apps, or when the user asks not to be interrupted. Enforces non-disruptive Computer Use: do not switch workspace/window, do not activate apps, and prefer app accessibility-tree element operations.
draw-technical-diagrams
Turn technical explanations into precise, publication-ready architecture diagrams, flowcharts, sequence diagrams, hub-and-spoke maps, side-by-side comparisons, token/cache diagrams, and technical covers. Use when Codex must decide what diagram to draw, derive a diagram brief from domain knowledge, match the spatial layout of a reference image without copying its content or style, generate SVG/PNG assets, or fix broken arrows, crowded routing, text overflow, and mobile readability.
engineering-contract-guard
Lightweight contract preflight for software engineering changes. Use when modifying or designing APIs, DTOs, request/response schemas, OpenAPI/Swagger docs, gateway routes, SDK/client contracts, storage/query isolation fields, or cross-service interface boundaries. This skill checks compatibility and current facts before coding; it is not a full implementation workflow.
iteration-drift-guard
通用工程迭代防跑偏守卫。用于长期迭代、迁移/拆分、连续修同类 bug、prototype 转 production、本地联调补丁、AI 反复生成代码、架构边界变模糊，或用户提到“跑偏”“越改越乱”“codex一直在写”“连续修 bug”“迁移”“重构”“联调”“fix after fix”“migration”“refactor”“local integration”“gone off track”等场景。它要求实现前做一次简短的边界、不变量、正式入口、归属和测试闭环检查，避免补丁堆叠成架构漂移。单纯 API/DTO/schema/Swagger/gateway 兼容性检查优先使用 engineering-contract-guard。
notebooklm
通过 `notebooklm` CLI 直接访问 Google NotebookLM。支持 notebook/source/chat/artifact 全链路操作，包括报告、闪卡、脑图、数据表生成与导出，以及 source fulltext 提取。
spark
Use when the user invokes spark, asks to use the Spark model, wants a fast subagent, or wants quick low-risk side work such as code search, lightweight investigation, summarization, or fact-finding. This skill treats the invocation as an explicit request to spawn a GPT-5.3 Codex Spark subagent and summarize its findings.
vercel-design
Use Vercel-inspired DESIGN.md guidance as the preferred visual baseline for frontend UI design and implementation. Use when Codex is designing or building web app interfaces, SaaS/developer-tool dashboards, developer-platform landing pages, or when the user mentions Vercel style, getdesign.md, DESIGN.md, monochrome precision, Geist, or Vercel-like UI. Prefer this guidance over weak, generic, or inconsistent existing UI styling unless the project has a clearly stronger brand system or the user says otherwise.
agent-skill-manager
Install, update, audit, and compare skills across Codex, Hermes, and Pi using the local skill registry and fixed sync script.