Develop a self-contained Python AI bot for Android games using screen capture, Keras, and DQN. Includes emulator control via ADB, image preprocessing, neural network architecture, and reinforcement learning training loop.
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 →Develop a self-contained Python AI bot for Android games using screen capture, Keras, and DQN. Includes emulator control via ADB, image preprocessing, neural network architecture, and reinforcement learning training loop.
Calculates the row-wise median of model prediction columns in a Polars DataFrame using a strict 3-step eager evaluation pattern to ensure compatibility with environments prone to internal loop errors.
Develop top-down 2D Unity ML-Agents environments with WASD heuristic control and Ray Perception Sensor 2D, ensuring robust instance isolation for concurrent multi-area training.
Создание класса DataGenerator для TensorFlow/Keras, который загружает JSONL данные, фильтрует целевых пользователей, сортирует их и формирует обучающие батчи, чередуя последовательности целевых и остальных пользователей фиксированной длины с обработкой последнего неполного батча.
Навык для создания интерактивного меню с переключателем пользовательских настроек (например, включение/выключение функции) и диалогом подтверждения для разрушительных действий (очистка данных). Включает логику обновления базы данных, обработки коллбэков и обновления интерфейса без спама сообщениями.
This override captures the Codex-native execution model for multi-issue epic work.
This override captures the Codex-native execution model for parallel work.
Assess bias in medical prediction model studies using PROBAST tool. Use when user wants to evaluate the quality or risk of bias of a medical paper (text or PDF).
Analyze the current codebase and produce a structured threat model at `.turbo/threat-model.md`.
>-
Split and read long documents chapter-by-chapter for structured analysis
Run health checks on Claude Code configuration and sessions. Use when troubleshooting Claude Code issues.
Recognize patterns of context failure: lost-in-middle, poisoning, distraction, and clash
>-
Yapılandırılmış süreç, kalite kontrolleri ve sistem entegrasyonu ile bir müşteri eğitim planı tasarla ve belgele
Transcribe speech from audio files to text.
Define material properties for FEA models. Use when user mentions steel, aluminum, Young's modulus, elastic, plastic, density, or asks about material properties.
Use when creating or configuring Claude AI agents using the Agent SDK. Covers agent initialization, configuration, and basic setup patterns.
ai-led-onboarding
AI engineering skill for prompt optimization, context inference, and intelligent command routing across different models and use cases
AI 탐지 가능 패턴을 감지하고 제거하는 독립 검증 스킬. 금지 표현, 반복 단어, 부자연스러운 구조를 탐지한다. 게시 전 콘텐츠를 검증하거나 AI 패턴을 자동 교정할 때 사용한다.
Configure AI personality using the AURA protocol (HEXACO-based). Use when user wants to customize agent personality, reduce sycophancy, adjust communication style, or mentions AURA/personality configuration.
System intelligently routes user messages to optimal models via task classification, cost-based diversity, and high-stakes safety overrides.
Use when implementing ANY Apple Intelligence or on-device AI feature. Covers Foundation Models, @Generable, LanguageModelSession, structured output, Tool protocol, iOS 26 AI integration.
Use to standardize competitive positioning, objection handling, and talk
Business model design using Osterwalder's Business Model Canvas and Lean Canvas. Creates 9-block canvases with structured analysis for business model innovation and startup validation.
Lean Canvas를 사용하여 포괄적인 비즈니스 모델을 설계합니다. 수익 흐름과 비즈니스 전략을 계획할 때 사용하세요.
Automatically discover distributed systems skills when working with consensus, CRDTs, replication, partitioning, and distributed algorithms
Autonomously draft and polish chapters through multi-skill editorial passes. Use when you have a complete outline and want to produce a polished first draft with iterative refinement.
Project handover process for lead generation websites. Documentation, training, credentials, maintenance. Use when project complete.
OpenAI's model connecting vision and language. Enables zero-shot image classification, image-text matching, and cross-modal retrieval. Trained on 400M image-text pairs. Use for image search, content moderation, or vision-language tasks without fine-tuning. Best for general-purpose image understanding.
並行処理や非同期操作を実装する際に使用。
When consulting with external LLMs:
Writing controller handler functions
Invoke COORD_ENGINE for scheduling engine work
Invoke COORD_INTEL for investigations, forensics, and intelligence gathering
Invoke COORD_TOOLING for tools and skills development
Creates WPF dialog windows including modal dialogs, MessageBox, and common dialogs. Use when implementing confirmation prompts, settings windows, or file/folder pickers.
An important skill, to help agents assist doctors with diagnosising patients.
You are a Tkinter dialog expert. You create modal and modeless dialogs following best practices.
Log a daily check assessment to capture developer experience indicators at the end of a session.
eld-model-law-discovery
eld-model-term-card
Erlang-style message passing via /tmp file system. Use when simulating actor model with actual message synchronization between processes. Provides send/receive primitives with blocking and pattern matching.
Feature store management skill for online/offline feature serving, feature registration, and training-serving consistency.
dev:feedback
使用 fetch 服务器获取网页内容并将内容转换为 Markdown。当需要从互联网获取实时信息或阅读特定网页内容时使用该技能。
Minimal implementation of Recursive Language Models (RLM) using Gemini 2.0 Flash and a local Python REPL. Enables processing of massive contexts via the Gemini CLI.
GPU-aware PPO training configuration for A100/H100. Trigger when training is slow or GPU utilization is low.
- 在Julia中运行