MLOps
模型部署、评估与运维
marketing-system-index
amogha-dalvi
Use when starting any marketing work for a SaaS business at 1M-5M ARR, when needing to find the right marketing skill, or when planning which marketing skills to execute and in what order.
designing-gtm-strategy
amogha-dalvi
Use when the user needs to translate foundational marketing insights into an operational growth plan, select a GTM motion, prioritize channels, or create a 90-day execution plan. Use when marketing efforts feel like disconnected tactics without a coherent system connecting them to revenue targets.
fuel
openclaw-rocks
"Optimized LLM inference and agent config for OpenClaw. Multi-provider routing with automatic cheapest-provider selection, context pruning, smart compaction, cheap heartbeats, session initialization, prompt caching, and memory management — all calibrated for autonomous agents that run for hours without wasting tokens. Triggers on: 'save on inference,' 'cheaper models,' 'optimize costs,' 'LLM config,' 'model routing,' 'inference setup,' 'fuel,' 'reduce token usage,' 'context management,' or any request to make an OpenClaw agent more cost-effective."
tracking-marketing-metrics
amogha-dalvi
Use when marketing lacks a metrics hierarchy connecting activity to business outcomes, when attribution is last-click or nonexistent, when marketing and sales report different numbers, when dashboards are ignored or full of vanity metrics, or when leading indicators are not tracked. Use when decisions are made on gut feel instead of data.
context-fundamentals
Nymbo
Understand the components, mechanics, and constraints of context in agent systems. Use when designing agent architectures, debugging context-related failures, or optimizing context usage.
flash-attention
tylertitsworth
"Flash Attention, FlashInfer, SDPA backends, PagedAttention, and attention kernel selection/configuration. Use when choosing or configuring attention backends for training or inference (FlashAttention-2/3, FlashInfer, SDPA, xFormers, PagedAttention, Ring Attention, FlexAttention/FlexDecoding, varlen_attn)."
aiconfigurator
tylertitsworth
"NVIDIA AIConfigurator — optimal LLM serving configuration for disaggregated/aggregated deployments, parallelism selection (TP/PP/EP/DP), quantization, and MOE planning. Use when planning model deployment topology on NVIDIA GPUs."
data-product-thinking
hollandkevint
First-principles reasoning for data product decisions. Frames problems as data products, not dashboards or pipelines. Use when evaluating data product strategy, making build-vs-buy decisions, scoping data product features, assessing product-market fit for data offerings, or when someone asks "should we build this data product?"
google-adk-python
vibery-studio
Build AI agents with Google's Agent Development Kit (ADK) Python - an open-source toolkit for building, evaluating, and deploying AI agents. Features LlmAgent, workflow agents (sequential, parallel, loop), tool integration, multi-agent systems, and deployment to Vertex AI or Cloud Run.
ethical-risk-assessment
hollandkevint
Ethical data risk evaluation, bias testing protocols, and governance practices for data products. Use when evaluating ML/AI features for fairness, designing bias testing protocols, planning phased rollouts for high-risk changes, reviewing data governance practices, or when someone asks "could this model be biased?" or "how do we ship AI features responsibly?" For HIPAA-specific guidance, see healthcare-data-domain.
kubeflow-trainer
tylertitsworth
"Kubeflow Trainer v2 — TrainJob CRD, Training Runtimes, Python SDK, JobSet, multi-framework support. Use when orchestrating distributed training on K8s. NOT for inference or Ray Train."
axolotl
tylertitsworth
"Axolotl — config-driven LLM fine-tuning framework. YAML-based SFT, instruction tuning, chat fine-tuning, DPO/IPO/KTO/ORPO preference optimization, GRPO reinforcement learning, reward modeling, LoRA/QLoRA, full fine-tuning with FSDP/DeepSpeed multi-GPU, N-D Parallelism (TP+CP+FSDP composition), multimodal VLM training, text diffusion training, QAT with NVFP4, sample packing, Flash Attention, dataset preprocessing, and checkpoint management. Use when fine-tuning or post-training LLMs with Axolotl."
kueue
tylertitsworth
"Kueue — ClusterQueues, ResourceFlavors, fair sharing, preemption, TAS, MultiKueue. Use when managing batch workload queuing and GPU quotas on K8s. NOT for Volcano."
llm-integration
liauw-media
"Use when integrating LLM APIs into applications. Covers API patterns, prompt templates, streaming, error handling, cost optimization, and provider abstraction. Apply when building chat interfaces, completion endpoints, or AI-powered features."
keda
tylertitsworth
"KEDA — ScaledObject/ScaledJob CRDs, 60+ triggers, scale-to-zero for GPU inference, TriggerAuthentication. Use when configuring event-driven pod autoscaling. NOT for HPA basics or Kueue."
swiftui-development
ProjAnvil
专注于使用 modern SwiftUI 构建用户界面，涵盖 NavigationStack, Observation 框架和 SwiftData 集成。
machine-learning
pluginagentmarketplace
Supervised & unsupervised learning, scikit-learn, XGBoost, model evaluation, feature engineering for production ML
wpf-mvvm-generator
JeongHeonK
WPF MVVM code generator. Creates ViewModel, View, and Model with CommunityToolkit.Mvvm source generators. Use when scaffolding new WPF features or entities following MVVM pattern.
Businessstrategy
robdtaylor
MBA-trained business strategist with full commercial awareness. USE WHEN user says 'business plan', 'financial analysis', 'P&L', 'cash flow', 'valuation', 'market analysis', 'competitive analysis', 'SWOT', 'business model', 'unit economics', 'fundraising', 'pitch deck', or needs strategic business guidance.
hytale-custom-assets
MnkyArts
Create and manage custom assets for Hytale including models, textures, sounds, particles, and asset packs. Use when asked to "add custom assets", "create textures", "make models", "add sounds", "configure particles", or "build an asset pack".
funsloth-runpod
chrisvoncsefalvay
Training manager for RunPod GPU instances - configure pods, launch training, monitor progress, retrieve checkpoints
Skillsmatrix
robdtaylor
Operator training and competency tracking for MNMUK machine shop. Supports IATF 16949 competency requirements and workforce planning. USE WHEN user asks 'who can run', 'training matrix', 'competency', 'skills matrix', 'operator certification', 'who is trained on', or workforce planning questions.
cloudrouter
manaflow-ai
Manage cloud development sandboxes with cloudrouter. Create, sync, and access remote VMs with GPU support, Docker, and browser automation. Use when asked to create a sandbox, spin up a dev environment, run code in the cloud, use GPUs, automate a browser, or interact with remote VMs.
workout-generator
LKB-99
"Generate personalized workout plans with progression and tracking. Use this skill when users want to create a workout routine, training program, or exercise plan. Triggers: workout, training, exercise, fitness plan, gym routine, treino, academia, plano de treino, rotina de exercícios."