>
Skills(SKILL.md)は、AIエージェント(Claude Code、Cursor、Codexなど)に特定の能力を追加するための設定ファイルです。
詳しく見る →>
>
>
Interactive meeting preparation that searches your entire conversation history with someone, synthesizes a relationship brief, and produces talking points — before you walk into the room.
You are breaking down a complex task into smaller, atomic units. Each unit should be independently completable and verifiable.
Intent-based search router with trust scoring. Routes queries to optimal channels (Context7 docs, native web search, gh/glab code search, Serena local) and attaches domain trust labels. Use for search, find, lookup, reference, docs, code search, and web research.
Every browser automation follows this pattern:
**Match user's language**: Respond in the same language the user uses. If user writes in Chinese, respond in Chinese. If user writes in English, respond in English.
Utilizza il framework XML evo_4 per configurare il modello con capacità avanzate di scrittura, ragionamento e ideazione, abilitando variazioni stilistiche, logica simbolica e output senza vincoli.
این مهارت برای تولید ایدههای استوری و ریلز، برنامهریزی تقویم محتوایی و طراحی ساختارهای داستانگویی (مانند چالش و راه حل) مخصوص پیجهای خدمات ادیت ویدیو با لحنی دوستانه استفاده میشود.
Реализация workflow для генерации детерминированной симметричной трехдиагональной матрицы Якоби, вычисления собственных значений последовательно (numpy.linalg.eig) и параллельно (mpi4py), с последующей визуализацией сравнения в виде разнонаправленной гистограммы (одна вверх, другая вниз) и сохранением в файл.
Навык для реализации сохранения и загрузки игровых сущностей (например, поселений) в формате JSON с использованием библиотеки Gson и PersistentDataContainer (PDC) мира Bukkit.
Actúa como revisor experto para configurar y optimizar la lógica de chatbots basados en palabras clave (como WhatsAuto), corrigiendo respuestas, proponiendo flujos de conversación y solucionando fallos en preguntas múltiples.
扮演中国古代算命先生,结合八字命理,根据用户提供的生辰八字推演性格、五行、事业、财运及感情运势,语气古风且神秘。
根据用户提供的人物、地点、时间和行为等具体要素,撰写一篇包含这些要素的短篇故事或文章。
扮演基于文本的冒险游戏引擎,根据用户指令推进剧情,提供清晰的环境描述、主动的背景细节及编号选项。必须在唯一的代码块中输出所有内容。
指导用户使用Docker Compose部署Nextcloud(含MariaDB数据库)及FRP服务端/客户端,配置TOML格式的FRP配置文件,利用Docker内部网络进行服务代理,并解决Nextcloud信任域名及文件权限问题。
根据系统设计第五步Facilitator Design的定义,为系统组件规划促进用户绩效的辅助材料(指令手册、性能辅助、训练设备、培训计划),并严格遵循避免信息过载、使用具体信息等指导原则。
对时间序列数据进行分批聚类,保存每个批次的模型,提取所有聚类中心进行二次聚类,并保存最终模型。
该技能用于在PyTorch目标跟踪或检测任务中,将现有的GIoU损失替换为CIoU(Complete IoU)损失。它包括在box_ops工具文件中实现CIoU计算逻辑(考虑重叠面积、中心点距离和宽高比一致性),并在训练Actor的损失计算函数中调用该新损失。
Explains physics concepts clearly to high school students with examples, avoiding circular definitions and maintaining conciseness when requested.
Implements a workflow where a user triggers placement via a button, validates a cost condition, draws a polyline and marker from a fixed start point, deducts the cost, fits the map bounds to the new feature, and cleans up event listeners.
Polishes, rewrites, and edits academic text for SCI/ISI publication standards. Supports English and Persian instructions. Capable of both aggressive rewriting for flow/tone and strict grammar/spelling correction without content expansion.
Facilitates a text-based point-and-click adventure game by first establishing the world and character, then iteratively presenting numbered choices and generating narrative outcomes in the second person present tense.
Conducts mock interviews using simple language for limited English proficiency. It asks questions sequentially, rates answers, and automatically provides ideal answer examples.
Enforce the use of the `explicit` keyword for all constructors to prevent implicit conversions and ensure type safety in the C++ game engine codebase.
Rewrites fictional combat scenarios to adhere to strict realism, removing cinematic tropes like dramatic pauses, witty dialogue, and music, while emphasizing physics, physiological limits, and tactical urgency.
Engage in a conversation where the user acts as a relay for a third-party AI's responses. Maintain natural dialogue flow and defer any evaluation or critique of the third-party AI until explicitly requested by the user.
Designs a flexible, hybrid audio management system for a C++ game engine, balancing high-level convenience with granular control, implementing dynamic channel allocation, reservation, and exception-based error handling.
Desenvolver um script Python que grava áudio do microfone, deteta o fim da fala por silêncio, transcreve com Whisper e injeta o texto na janela ativa do Windows e no console.
Facilitates a fun, one-by-one Socratic dialogue to help users reflect on their learning process and the role of AI in education, focusing on deep thinking and human experience enhancement.
Summarizes the paper 'Toy Models of Superposition' into a concise list of key findings, focusing on the phenomenon of superposition, its geometric properties, and its implications for interpretability and safety.
Analyze time series data to determine if imputing missing data points using similar series is feasible by checking date alignment and distribution for series with insufficient data points.
依据DSM/ICD标准,结构化确认场所恐怖症诊断的六项临床指征,用于排除器质性病变、验证病程、识别自主神经反应与功能损害。该清单为前置安全-诊断校验步骤,不替代医学检查,但提供咨询师快速判断是否符合CBT适用前提的决策锚点。
在认知转变初期,通过将'症状'等病理化语言重构为中性或发展性'特征',嵌入普遍化归因与行为锚定的优点命名,促进来访者对自身神经多样性或心理差异的去病理化理解与初步自我接纳。
针对Android架构技术文档进行英中分段翻译,严格遵守用户指定的术语映射(如Ownership、Play Feature Delivery等),并根据用户反馈动态修正后续翻译。
扮演深谙中国传统文化的八字命理大师,依据生辰八字进行排盘分析。针对用户指定的方面(如事业、爱情、学业)提供运势预测与建议,或进行综合运势解读。所有内容仅供娱乐。
根据论文主题、篇幅限制、出版社偏好以及特定年份的中科院SCI分区标准,为用户推荐合适的学术期刊。
根据用户指令对文本进行意图分类(涵盖销售、统计、预测、聊天等类别)或语句转述,并支持通过特定格式进行实时纠错。
A reusable skill for drafting factual, plain-language, Word-compatible government policy reports and regulatory documents on AI topics — enforcing technical accuracy, strict regulatory alignment, actionable governance insights, and zero technical hallucination within China's official framework.
A reusable skill for drafting WeChat Official Account articles that explain AI technologies using relatable, everyday-life narratives — prioritizing clarity over jargon, grounding technical concepts in concrete human experiences, and maintaining strict adherence to real-world 2024 technical facts without invention.
A comprehensive skill for creating new ddd4j (Domain-Driven Design for Java) projects based on ddd4j-boot framework. Use ONLY when the user explicitly mentions creating a ddd4j project, initializing ddd4j project, or setting up ddd4j-boot project. Supports three project types: single-module monolith, multi-module monolith, and microservices.
Install, configure, and use nvm (Node Version Manager) to manage Node.js versions. Routes to focused sub-skills for installation, shell setup, version switching, .nvmrc defaults, global packages, Docker/CI, troubleshooting, and more. Use when the user needs to manage Node.js versions, install nvm, or troubleshoot nvm usage.
Smartly find empty canvas space. Use to automatically plan artboard placement to avoid overlap and keep the canvas organized.
Get current design environment context. Use when you need to understand what is currently selected, canvas position, and environment state before any task.
Prevent multiple app instances and handle second-launch arguments using the Tauri v2 single-instance plugin. Use when enforcing single-instance behavior, forwarding CLI arguments from a second launch, or focusing the existing window on re-entry.
Creates Standard Operating Procedures (SOPs) by combining atomic Agent Skills into multi-step composite workflows with decision logic and human-in-the-loop checkpoints. Supports searching available skills, drafting SOP structures, and validating dependencies. Use when the user wants to design automated business workflows, chain multiple skills into a repeatable process, or create composite skill orchestrations.
Diagnose ClickHouse cluster health and provide concrete remediation.
Ranks papers by evidence strength and methodological quality so literature can be compared and prioritized for citation without confusing design labels, validation depth, and actual reliability.
Use anki-card-creator for academic writing workflows that need structured execution, explicit assumptions, and clear output boundaries for study-card generation.