Box-Agent

景观施工图预审专家级 Skill「总工之眼」v3.1。基于李博总工 20 年实战经验，集成中国规范 1500+ 条、12 条经总工终审与五专业外部工程师复核的经验法则库、三级签字判断逻辑。**任何涉及景观图纸审查、施工图预审、图纸会审、规范核查、绿地率/无障碍/坡度/植物配置/海绵设施/雨水花园/景观照明/亭廊架/无障碍流线/竖向标高/铺装做法的场景，都应使用本技能。** 触发关键词包括：图纸审查、施工图审、方案审查、总工审图、签字建议、规范核查、绿地率、无障碍、坡度、种植间距、覆土深度、海绵城市、雨水花园、景观照明、亭廊架基础、跨专业会签、北方公园、冬季结冰、地下室顶板大乔木、设计院预审、设计院总工、景观合规、施工图错漏、漏审、暂缓签字、不可签字、整改通知、审图档案、landscape review、施工图、构筑物、给排水、铺装。即使用户没有明确说"审查"，只要提到景观图纸、施工图、规范、签字、整改等场景，也应该使用本技能。

Create, inspect, edit, validate, render, and QA presentation decks. Use when the user mentions PowerPoint, PPT, PPTX, HTML deck, slide deck, presentation, template slides, speaker notes, slide images, or asks to read, generate, create, make, design, or modify a presentation artifact. New decks default to controlled, editable HTML delivery; PPTX is an explicit optional export.

HyperFrames MP4/GIF videos; motion graphics; explainer clips; 视频、短片、动图、动画成片。

Evidence-first research synthesis workflow for market research, industry analysis, competitive research, policy or company information synthesis, file-only or file-augmented analysis, technical architecture reviews, risk assessment, and report-ready conclusions. Use when the user asks for comprehensive, cross-verified research or deep summarization rather than a simple lookup.

Toolkit for styling non-deck artifacts with one of 10 legacy color/font themes. PPT, PPTX, slide, and deck theme discovery is owned by the pptx skill and must be redirected there.

Guides the agent to proactively save user information to long-term memory and search context when needed. Load this skill to understand when and how to use memory_write, memory_read, and memory_search.

生成单文件 HTML 数据看板。触发：用户要把数据/分析结果做成网页、dashboard、可视化报告、management demo。

Use when generating HTML pages, slide decks, posters, or reports that should adopt a specific visual style — whether invoked directly by the user OR called by another skill (e.g. pptx, poster, report skills) that has already decided the structure but needs to pick and apply a visual identity. Matches the brief's mood / industry / density against 32 pre-extracted design "Visual DNA" profiles, then injects the chosen profile's color, typography, and decoration rules as hard constraints into the generation prompt. The caller writes HTML inside their own structural conventions; this skill only governs the look.

管理 MCP 服务器配置（mcp.json）：查看、添加、移除、启用、禁用

面向企业 IP、品牌吉祥物、球迷社群角色和活动主视觉的合规海报生成技能，将用户心中的球星英雄记忆转译为原创 IP、抽象精神符号和可发布的海报提示词，并提供跨平台叠字、字体 fallback、错误码和用户排障引导。

扫描本地已安装技能，稳定解释技能用途和唤起方式；根据用户任务推荐单技能或多技能组合；本地无匹配时，给出 SkillHub 搜索、筛选、确认安装与启动引导；同时提供健康检查、异常恢复和 FAQ。

Create beautiful visual art in .png and .pdf documents using design philosophy. You should use this skill when the user asks to create a poster, piece of art, design, or other static piece. Create original visual designs, never copying existing artists' work to avoid copyright violations.

Convert deep research, long-form source text, reports, notes, or collected evidence into a PPT/deck outline that can be progressively expanded. Use when the user wants a standalone outline, page-by-page deck structure, speaking-prompt outline, or full speaker-script outline from research material. Do not use for creating PPTX files, visual design, slide rendering, or template layout work.

Use this skill whenever the user asks to create, export, save, append, prepend, open, or write Markdown notes in Obsidian. It explains when to call the native obsidian_* tools backed by the official Obsidian CLI, and how to avoid unsafe direct bash CLI writes.

针对用户点名的某一场世界杯对阵，结合内置 48 强球队画像、评分模型与实时信息（上轮结果/伤停/首发/状态），用高精度对位引擎算出比分、胜平负、置信度与进球线索。适用于“世界杯比分预测”“预测某场比赛”“谁会赢”“怎么进球”“胜负预测”“对阵预测”等请求；只做单场预测，不做未来多场批量。触发词：世界杯、世界杯预测、比分预测、预测某场比赛、谁会赢、进球方式。

Helps the user create a recurring or one-off scheduled task through conversation. Load this skill whenever the user wants something done on a schedule (every day / every week / a specific time / "remind me" / periodic reports or monitoring). It explains how to confirm requirements first, translate them into a cron expression, and then call the create_scheduled_task tool to pop the pre-filled creation window.

自媒体接广制作与复盘向导。适配办公小浣熊安装使用，带新人博主一步步完成接广内容制作和复盘：收集 brief、拆解需求、校准卖点、联网搜索、生成选题、打磨脚本、拆拍摄发布任务、交付检查、发布后数据复盘。 触发方式：/self-media-ad-workflow、/自媒体接广、「第一次接广」「接广告」「商单脚本」「推广 brief」「帮我做接广流程」「复盘接广内容」

Generate high-potential viral title candidates for content across WeChat public account articles, X/Twitter posts, YouTube videos, Bilibili videos, and similar content platforms. Use when the user needs headline/title ideation, batch title generation, platform-specific title adaptation, title-library reuse, title selection, title scoring, or a reusable title workflow based on universal formulas plus separated platform methods such as 公众号, X, YouTube, and B站.

Route viral topic discovery across platform-specific account-growth skills. Use when the user asks for 起号选题, low-follower viral references, recent viral content, cross-platform topic mining, or does not specify whether to search WeChat, X, Bilibili, or YouTube.

用于生成世界杯战报简报、焦点战预告、球队/球员动态，以及按用户偏好定制的世界杯内容。默认中文输出、球迷向语气，每篇简报与预告都必须包含关键人员信息；简报默认只覆盖昨日比赛、最多近 3 天，并产出单页 HTML 报告。适用于“生成世界杯简报”“写一份焦点战预告”“今天世界杯有什么重点比赛”“重点关注某队/某球员”“更新世界杯关注偏好”等请求。触发词：世界杯简报、世界杯战报、焦点战预告、比赛预告、世界杯重点比赛、关注偏好。

Design skill architecture before creation. Use when the user wants to discuss whether an ability should become a single skill, a skill suite, an expert, an expert team, or a hybrid pipeline; produce a structured architecture plan and, when appropriate, an Office Raccoon skill-blueprint handoff for the Skills page.

Create and refine Codex skill files after the skill architecture is clear. Use when writing or editing SKILL.md, references, scripts, templates, trigger wording, or validating that an existing skill package is concise, discoverable, and portable across macOS, Windows, and Linux. For deciding single skill vs skill suite vs expert/team architecture first, use skill-architect.

Prepare isolated git branches or worktrees for software-development tasks. Use before larger coding work when a clean branch, worktree, baseline verification, or separation from existing user changes is needed.

Review software changes for correctness, regressions, missing tests, maintainability, security, and behavioral mismatches. Use when asked to review code, before merging, after implementing a major task, or when a fresh risk-focused pass is needed.

Systematically debug software failures before proposing fixes. Use for test failures, runtime errors, build failures, flaky behavior, performance regressions, integration failures, logs, screenshots, or any unexpected technical behavior where root cause must be established from evidence.

Execute an existing software implementation plan task-by-task. Use when a written plan, checklist, issue breakdown, or approved spec should be implemented with incremental progress, local verification, and careful handling of blockers.

Finish a software-development branch or completed coding task. Use after implementation and verification to summarize changes, inspect git status, prepare commit or PR notes, choose merge/push/keep options, and avoid accidentally staging unrelated user changes.

Create concrete implementation plans for software-development tasks after requirements are known. Use when a feature, bugfix, refactor, migration, or integration needs file-level steps, sequencing, test strategy, risk notes, and verification before coding.

Handle code review feedback carefully. Use when review comments, requested changes, PR feedback, CI review output, or human reviewer notes need to be understood, verified against the codebase, implemented, or pushed back on with technical reasoning.

Turn rough software-development requests into clear, reviewable specifications before implementation. Use for new features, behavior changes, ambiguous bugfix scope, product requirements, technical designs, acceptance criteria, or when a coding task needs agreement on intent, constraints, and non-goals before code changes.

Apply test-driven development to software features, bugfixes, refactors, and behavior changes. Use before writing implementation code when the behavior can be tested, especially for shared logic, regressions, parsers, state transitions, APIs, and UI behavior with existing test harnesses.

Verify software work before claiming it is complete. Use before saying a fix works, tests pass, a build is clean, a bug is resolved, a PR is ready, or a task is done; requires fresh command output, manual checks, or explicit skipped-verification notes.

写作工作流：为写作专家、专家团或普通写作任务提供选题、大纲、初稿、润色、风格参考、素材成文、标题优化、事实核查、多平台改写和发布前质量门禁。适用于把模糊想法、零散素材、已有文本、用户授权知识库、阅读摘录或公开资料，转化为结构清晰、观点鲜明、风格统一、事实可靠、适合交付或发布的内容。

Suite of tools for creating elaborate, multi-component claude.ai HTML artifacts using modern frontend web technologies (React, Tailwind CSS, shadcn/ui). Use for complex artifacts requiring state management, routing, or shadcn/ui components - not for simple single-file HTML/JSX artifacts.

Comprehensive document creation, editing, and analysis with support for tracked changes, comments, formatting preservation, and text extraction. When Claude needs to work with professional documents (.docx files) for: (1) Creating new documents, (2) Modifying or editing content, (3) Working with tracked changes, (4) Adding comments, or any other document tasks

Comprehensive PDF manipulation toolkit for extracting text and tables, creating new PDFs, merging/splitting documents, and handling forms. When Claude needs to fill in a PDF form or programmatically process, generate, or analyze PDF documents at scale.

Comprehensive spreadsheet creation, editing, and analysis with support for formulas, formatting, data analysis, and visualization. When Claude needs to work with spreadsheets (.xlsx, .xlsm, .csv, .tsv, etc) for: (1) Creating new spreadsheets with formulas and formatting, (2) Reading or analyzing data, (3) Modify existing spreadsheets while preserving formulas, (4) Data analysis and visualization in spreadsheets, or (5) Recalculating formulas

Use when the user asks to send, draft, reply to, forward, search, read, summarize, organize, or triage email through a generic webmail UI when no provider-specific mail tool is available. Uses Playwright/browser automation to open webmail, fill recipients/subject/body, inspect inboxes, and perform explicit user-authorized sends.

A set of resources to help me write all kinds of internal communications, using the formats that my company likes to use. Claude should use this skill whenever asked to write some sort of internal communications (status reports, leadership updates, 3P updates, company newsletters, FAQs, incident reports, project updates, etc.).

Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.