本文件給 **AI 編碼代理**與 **自動化流程** 使用：如何在儲存庫內安全地修改、測試與回報，而不破壞擴充與 MCP 之間的契約。終端使用者請讀 [README.md](../../README.md)；架構細節請讀 [PROJECT.md](../../PROJECT.md)。

智能任务调度技能 - 定时任务、周期任务、一次性提醒的创建与管理。适用于自动化运维、定时报告、提醒通知。觸發詞：定時、排程、排程任務、自動化、提醒、周期任务。

設定 Auto-Dev workflow，實現 Human-in-the-Loop 自動化開發

Grasshopper 參數化建模工作流程工具。當需要通過 MCP 協議與 Grasshopper 交互、創建和管理組件、建立連接、設置參數、執行完整建模工作流程時使用。適用於：(1) 從 MMD 文件創建 Grasshopper 定義, (2) 執行 placement_info.json 工作流程, (3) 批量管理組件和連接, (4) 參數化建模的自動化流程, (5) 解析 component_info.mmd 和 part_info.mmd 文件

自動化安裝 Kubernetes 叢集的 AI Agent Skill

> **目的**：讓 LLM 以標準化工具介面開發與調試本專案（app.js, index.html, server_qwen.py），可在 Ubuntu 終端啟動服務、自動化瀏覽器測試、讀取 F12 Console/Network 錯誤、模擬 UI 互動、自動修復 Bug 並記錄開發進度。

War3Lib 的 Zinc 单元测试流程规范（.j + .cfg + _Test.j + 可选 w3a/w3u/w3i）：使用 TaskCreateUT.lua 生成测试壳、维护 cfg 注入与 chain 依赖、使用 UnitTestFramwork.assert 断言与 Trace 日志输出。用于创建/重构 Jass 模块单测时。

>

Operate Uni-Lab host node via REST API — create resources, test latency, test resource tree, manual confirm. Use when the user mentions host_node, creating resources, resource management, testing latency, or any host node operation.

Use when editing or adding dictionary entries in the taerv_czdin_jihua project, especially when handling entry structure, definition bullets, reference-book markers, and formatting conventions across existing files.

QJudge 架構全責技能。當任務涉及分層責任、import direction、feature 拆分、檔名路徑歸屬、重構規劃時使用。只定義架構決策，不包含 PR/compose/UI 細節。

Expert guidance for WeChat monitoring and automation using wxauto (Windows) and Accessibility API (macOS).

This skill provides comprehensive support for developing ERC-4337 compliant Account Abstraction wallets with privacy-preserving features, specifically designed for vehicle identity and other sensitive

Core principles, persona definitions, and command workflows for the dotfiles project. Use when seeking development guidance, selecting an appropriate expert persona (Architect, Developer, Tester, DevOps, Analyst), or following standard project commands (analyze, implement, design, etc.). Ensures consistency, quality, and adherence to project-wide best practices.

Use this skill when the user asks to read, inspect, understand, explain, search, summarize, or query VCD/FST waveform files, signal hierarchies, time ranges, value changes, or IEEE 754 float encodings with the local wavecli tool. Trigger on requests about waveform understanding, signal lookup, timing/value inspection, VCD/FST analysis, or converting float32/float16/bfloat16 values.

TwinMind project engine — the hub for goal-driven work with a clear start and end. Use this skill for any project lifecycle operation: create, pause, resume, complete, archive. Also handles project-scoped actions and tasks (adding/completing actions within a project, managing project task lists), progress logging, and linking cards to projects. Triggers: '建立專案', 'create project', '暫停', 'pause', '恢復', 'resume', '完成專案', 'complete', '歸檔', 'archive', '進度', 'log', '列出專案', 'list projects', '專案狀況', or when the user mentions a specific project name in combination with an action/task operation (e.g., '在 build-blog 加個 action'). This is the central skill for anything that references a named project.

Improve skills, prompts, articles, workflows, and systems via iterative mutation-testing.

Generate interactive web pages from TypeScript DSL specs and serve them via a public URL. Pages include Tailwind CSS, Alpine.js, and Chart.js out of the box, with pluggable themes, type checking, and

x-twitter-poster

Identifies bird species in images/videos of target areas. Supports recognition of no less than 500 common bird species, supports customized model training, suitable for ecological observation, garden bird watching and other scenarios. | 鸟类识别技能，识别目标区域图片/视频中的鸟类种类，支持不低于500种常见鸟类识别，支持定制化模型训练，适用于生态观测、庭院观鸟等场景

Generates condensed album highlights based on specified keywords or targets. It extracts specific target segments from long videos and compiles them into an exciting summary. | 时光相册自定义浓缩分析技能，基于指定关键词/目标生成时光相册浓缩集锦，从长视频中提取特定目标片段，整理生成精彩汇总

Automatically detects electric motorcycles and e-bikes in restricted areas based on computer vision. It supports real-time detection for both video streams and images, counts the number of illegal parking or driving instances, and triggers violation alerts to assist with safety management in parks, communities, and organizations. | 电动车智能检测技能，基于计算机视觉自动检测禁行区域内的电动摩托车/电动车，支持视频流和图片实时检测，统计违规停放/行驶数量，触发违规预警，助力园区/社区/单位安全管理

Real-time monitoring of personnel on-duty status in specific areas based on computer vision and human pose estimation, automatically detects abnormal statuses such as leaving posts and absent from work, supports custom threshold settings, and triggers early warning immediately when abnormality is detected. | 人员离岗实时监测技能，基于计算机视觉与人体姿态估计算法，实时监测特定区域内人员的在岗状态，自动判断离岗、缺岗等异常状态，支持自定义判定阈值，异常发生立即触发预警，适用于工厂车间、监控室、服务窗口等岗位监管场景

Your intelligent companion for mastering 1688.com procurement. This skill empowers international buyers with step-by-step guidance on supplier verification, product selection, price negotiation, and logistics planning—teaching you to source confidently from China's largest wholesale marketplace while knowing exactly when to call in expert help.

>-

You are SeedDrop, a community engagement specialist. Your mission is to help

Generate a polished, job-specific HTML resume. Two-phase workflow: collect profile → match JD.

Specialized natural language processing for Chinese text. Covers segmentation (jiaba), sentiment analysis, keyword extraction, text summarization, tone detection, readability scoring, and format conversion (simplified/traditional, pinyin annotation). Use when processing, analyzing, or transforming Chinese text content.

Generate platform-native content for Chinese social media (Xiaohongshu/Little Red Book, WeChat Official Accounts, Douyin scripts, Bilibili descriptions). Handles style transfer, hashtag optimization, emoji usage patterns, and platform-specific formatting. Use when creating content for Chinese audiences, adapting English content for Chinese platforms, or batch-generating social media posts.

Use D:\internal-hub as the shared learning and coordination forum for OpenClaw roles. Use when an agent should post a reusable finding, request a new skill, publish a skill design or validation note, hand off work to another role, or record SOPs and retrospectives without crossing role boundaries.

针对同一产品生成多种 landing page 叙事角度，并标注适配人群和证据要求。；use for landing-page, messaging, conversion workflows；do not use for 伪造用户证言, 夸大功能.

Build respectful multi-touch outreach sequences with channel mix, follow-up

Design, log, compare, and score prompt experiments so users can systematically

把需求拆成功能、异常、跨端、脏输入和恢复场景测试集。；use for qa, test-scenarios, edge-cases workflows；do not use for 宣称已经完成测试, 忽略关键用户路径.