MLOps
Operacions d’aprenentatge automàtic
visionos-design-guidelines
dirnbauer
Apple Human Interface Guidelines for Apple Vision Pro. Use when building spatial computing apps, implementing eye/hand input, or designing immersive experiences. Triggers on tasks involving visionOS, RealityKit, spatial UI, or mixed reality.
code-security-audit
ssrsec
AI 驱动的生产级代码安全审计协议。面向企业内部授权审计，默认使用结构化 harness：项目画像、攻击面枚举、Sink/Control 双轨候选发现、验证队列、能力链、质量门、报告渲染和完整证据交付。支持 PHP、Java、C# / ASP.NET、Go、Python、JavaScript / TypeScript Node.js、Ruby、Rust、Kotlin、Deno/Bun，以及 JVM/.NET 编译产物反编译计划。用户说「开始审计 / 对 XXX 做安全审计 / 代码审计 / 安全扫描 / 找漏洞 / security audit / 红队审计」时触发。
ML Training Debugger - Diagnose and Fix Training Issues
DNYoussef
Context Cascade - Nested Plugin Architecture for Claude Code Official Claude Code Plugin Version 3.1.0 Last updated: 2026-01-09 (see docs/COMPONENT-COUNTS.json for source counts) Context-saving nested architecture: Playbooks -> Skills -> Agents -> Commands. Load only what you need, saving 90%+ context space.
AppFactory Builder Skill
0xAxiom
Built by @AxiomBot and @meltedmindz. Ship something.
skills
atxinsky
Execute plan in batches with review checkpoints
skills
atxinsky
Create detailed implementation plan with bite-sized tasks
skills
atxinsky
"You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores requirements and design before implementation."
skill-commons
choru-k
Shared contract nodes reused by multiple skills (routing, delegation, lifecycle, and output conventions).
ipynb-notebooks
M4n5ter
面向 .ipynb Notebook（Jupyter / JupyterLab / Google Colab / VS Code）的创建、审阅、重构与展示。涵盖工程化目录结构、token 高效处理、演示/分享模式、以及 uv/venv 可复现工作流。
pytorch-to-numpy
spcl
Translate a KernelBench PyTorch model into numpy -- the torch-specific step in front of the python-to-numpy contract, which carries everything else.
ml-engineer
k1lgor
Use this for building machine learning models, feature engineering, training pipelines, and integrating predictions into applications.
pipeline-metrics
OmniNode-ai
Report pipeline health metrics — rework ratio, cycle time, CI stability, and feature velocity
executing-plans
OmniNode-ai
Use when partner provides a complete implementation plan to execute — reviews the plan critically, verifies live PR state against plan assumptions, creates Linear tickets via plan-to-tickets, then routes to epic-team (≥3 tickets) or ticket-pipeline (1-2 tickets)
aio-mental-models
aiocean
This skill should be used when the user faces complex decisions, problem-solving, debugging, system design, strategic thinking, or needs structured reasoning. Comprehensive mental models framework with 50+ models covering first principles, second-order thinking, inversion, feedback loops, and more.
model-researcher
IgorWarzocha
Add new/custom AI models to opencode.json. Use proactively for bleeding-edge releases, non-standard providers, self-hosted models, or custom endpoints. Examples: - user: "Add the new Claude 4.5" → websearch for API specs, add provider entry with baseUrl, verify model ID format - user: "Use my local Ollama instance" → configure custom provider with http://localhost:11434, set model ID format - user: "Configure this OpenAI-compatible proxy" → add provider with custom baseUrl, set apikey env var, verify compatibility - user: "Model X just released, add it" → research provider documentation, find model ID and capabilities, add to config
opencode-config
IgorWarzocha
Edit opencode.json, AGENTS.md, and config files. Use proactively for provider setup, permission changes, model config, formatter rules, or environment variables. Examples: - user: "Add Anthropic as a provider" → edit opencode.json providers, add API key baseEnv var, verify with opencode run test - user: "Restrict this agent's permissions" → add permission block to agent config, set deny/allow for tools/fileAccess - user: "Set GPT-5 as default model" → edit global or agent-level model preference, verify model name format - user: "Disable gofmt formatter" → edit formatters section, set languages.gofmt.enabled = false
grey-haven-evaluation
greyhaven-ai
"Evaluate LLM outputs with multi-dimensional rubrics, handle non-determinism, and implement LLM-as-judge patterns. Essential for production LLM systems. Use when testing prompts, validating outputs, comparing models, or when user mentions 'evaluation', 'testing LLM', 'rubric', 'LLM-as-judge', 'output quality', 'prompt testing', or 'model comparison'."
plan-down
VCnoC
Method clarity-driven planning workflow using zen-mcp tools (chat, planner, consensus). Phase 0 uses chat to judge if user provides clear implementation method. Four execution paths based on automation_mode × method clarity - Interactive/Automatic × Clear/Unclear. All paths converge at planner for task decomposition. Produces complete plan.md file. Use when user requests "create a plan", "generate plan.md", "use planner for planning", "help me with task decomposition", or similar planning tasks.
reinforcement-learning
Aznatkoiny
Reinforcement Learning best practices for Python using modern libraries (Stable-Baselines3, RLlib, Gymnasium). Use when: - Implementing RL algorithms (PPO, SAC, DQN, TD3, A2C) - Creating custom Gymnasium environments - Training, debugging, or evaluating RL agents - Setting up hyperparameter tuning for RL - Deploying RL models to production
deep-learning
Aznatkoiny
"Comprehensive guide for Deep Learning with Keras 3 (Multi-Backend: JAX, TensorFlow, PyTorch). Use when building neural networks, CNNs for computer vision, RNNs/Transformers for NLP, time series forecasting, or generative models (VAEs, GANs). Covers model building (Sequential/Functional/Subclassing APIs), custom training loops, data augmentation, transfer learning, and production best practices."
consulting-frameworks
Aznatkoiny
Core consulting thinking frameworks and methodologies for structuring business problems, communicating findings, analyzing strategy, building financial models, and designing operations. Use when any agent or command needs to apply MECE decomposition, pyramid principle, hypothesis-driven analysis, issue trees, SCR communication, Porter's Five Forces, TAM/SAM/SOM market sizing, value chain analysis, NPV/IRR decision criteria, build/buy/partner evaluation, RACI matrices, or any standard consulting framework. This skill provides procedural guidance — not just framework names, but how to apply them correctly.
nano-banana-pro
EnzeD
Image generation and editing using Google Gemini's Nano Banana Pro (gemini-3-pro-image-preview) model. Use when user requests: "Generate an image", "Create an image", "Make me a picture", "Draw", "Edit that image", "Change the color", "Remove background", "Add transparency", "Modify this image", "Make it transparent", "Change the style", "Add text to image", or any image creation/manipulation task. Supports text-to-image generation, image editing, multi-turn conversations, and transparency extraction via difference matting technique.
openclaw-setup
Aznatkoiny
Set up, install, configure, and deploy OpenClaw (formerly ClawdBot/MoltBot) — a personal AI assistant that runs on your own devices and connects to messaging channels. Use when users ask to "set up OpenClaw," "install ClawdBot," "install MoltBot," "deploy a personal AI assistant," "configure OpenClaw on Mac," "deploy OpenClaw to VPS," "set up OpenClaw on Hostinger," "connect OpenClaw to Telegram," "configure iMessage with OpenClaw," or any variation involving OpenClaw installation, gateway configuration, channel setup, Anthropic auth, or security hardening. Also triggers on "openclaw onboard," "openclaw doctor," "openclaw security audit," troubleshooting OpenClaw deployments, OpenClaw security, OpenClaw cost control, or ClawHub skills safety.
external-research-implications
TinyAssets
Research an external project, repository, paper, article, or codebase and derive TinyAssets implications, plans, and integration slices. Use when the user points an agent at outside work and asks what we should learn, copy, avoid, integrate, or build from it.