MLOps
Operacions d’aprenentatge automàtic
llm-inference
eyadsibai
Use when "LLM inference", "serving LLM", "vLLM", "llama.cpp", "GGUF", "text generation", "model serving", "inference optimization", "KV cache", "continuous batching", "speculative decoding", "local LLM", "CPU inference"
time-series
pluginagentmarketplace
ARIMA, SARIMA, Prophet, trend analysis, seasonality detection, anomaly detection, and forecasting methods. Use for time-based predictions, demand forecasting, or temporal pattern analysis.
systematic-debugging
ayeo-io
"Debugging from evidence: read the failure, form one hypothesis, test it. Load on an unexpected failure."
ml-engineering
eyadsibai
Use when "deploying ML models", "MLOps", "model serving", "feature stores", "model monitoring", or asking about "PyTorch deployment", "TensorFlow production", "RAG systems", "LLM integration", "ML infrastructure"
ai-avatar-video
alfredang
Create AI avatar and talking head videos with OmniHuman, Fabric, PixVerse via inference.sh CLI. Models: OmniHuman 1.5, OmniHuman 1.0, Fabric 1.0, PixVerse Lipsync. Capabilities: audio-driven avatars, lipsync videos, talking head generation, virtual presenters. Use for: AI presenters, explainer videos, virtual influencers, dubbing, marketing videos. Triggers: ai avatar, talking head, lipsync, avatar video, virtual presenter, ai spokesperson, audio driven video, heygen alternative, synthesia alternative, talking avatar, lip sync, video avatar, ai presenter, digital human
runpod-serverless-builder
AvivK5498
Build production-ready RunPod serverless endpoints with optimized cold start times. Use when creating or modifying RunPod serverless workers for (1) vLLM-based LLM inference, (2) ComfyUI image/video generation, or (3) custom Python inference. Supports both baked models (fastest cold starts) and dynamic loading (shared models). Generates complete projects including Dockerfiles, worker handlers, startup scripts, and configuration optimized for minimal cold start latency.
process-modeling
spjoshis
Master process modeling with BPMN, flowcharts, swimlane diagrams, and process optimization techniques for business process improvement.
slime-rl-training
magnetoid
Provides guidance for LLM post-training with RL using slime, a Megatron+SGLang framework. Use when training GLM models, implementing custom data generation workflows, or needing tight Megatron-LM integration for RL scaling.
statistical-analysis
spjoshis
Master statistical analysis with hypothesis testing, A/B testing, regression, and statistical methods for data-driven decisions.
concept-cartographer
PrakharMNNIT
"Generate visual concept maps, flowcharts, architecture diagrams, and relationship diagrams from structured notes or technical content using Mermaid syntax. Use when the user has lecture notes, study materials, or technical documentation and wants visual diagrams to aid understanding. Produces multiple diagram types: concept hierarchy maps, process flowcharts, architecture diagrams, comparison matrices, timeline diagrams, and mind maps. Trigger phrases: 'create diagrams from notes', 'visualize concepts', 'concept map', 'make flowcharts', 'diagram this', 'visual notes'."
transcript-pipeline
PrakharMNNIT
This skill should be used when the user asks to "process this transcript", "convert lecture to notes", "run transcript pipeline", "generate class tutorial from Zoom captions", "validate transcript coverage", or "enrich class resources" (Notion/Canva/Drive links) for bootcamp notes.
douyin-downloader
MgOxides
Use when needing to download and transcribe Douyin/抖音 videos to text. Supports single video URLs (/video/xxx) and user profile URLs (/user/xxx) with auto-detection. Triggers on Douyin links, 抖音 transcription requests, Chinese short video audio extraction, or batch video downloading.
machine-learning
pluginagentmarketplace
Supervised/unsupervised learning, model selection, evaluation, and scikit-learn. Use for building classification, regression, or clustering models.
go-concurrency
spjoshis
Master Go concurrency with goroutines, channels, select, sync primitives, and patterns for building concurrent applications.
@tank/opencode-agent-creator
tankpkg
Create specialized OpenCode agents that assume specific roles — frontend architect, DevOps SRE, database specialist, code reviewer, or any domain expert. Covers agent anatomy (markdown and JSON formats), role design (persona, expertise, behavioral directives), system prompt engineering, tool permissions, oh-my-opencode integration (categories, skill injection, delegation wiring), and converting existing Tank skills into standalone agents. Synthesizes OpenCode source (sst/opencode), oh-my-opencode v3.x agent system, and production agent analysis. Trigger phrases: "create agent", "opencode agent", "new agent", "convert skill to agent", "make an agent", "agent from skill", "custom agent", "specialized agent", "agent template", "agent for", "agent configuration", "agent role", "create a specialist", "build agent", ".opencode/agent"
backend-principle-eng-python-ml-pro-max
PrakharMNNIT
"Principal backend engineering intelligence for Python AI/ML systems. Actions: plan, design, build, implement, review, fix, optimize, refactor, debug, secure, scale ML services and pipelines. Focus: data quality, reproducibility, reliability, performance, security, observability, model evaluation, MLOps."
agentform
AvivK5498
Create and debug Agentform AI agent configurations (.af files). Use when: (1) Creating new agentform projects or workflows (2) Debugging agentform syntax errors (3) Adding MCP server integrations (4) Configuring agents, models, policies, or capabilities (5) Writing workflow steps with routing and human approval Agentform is "Infrastructure as Code for AI agents" - declarative .af files define agents, workflows, and policies.
voice-processing
SpiralCloudOmega
AI-powered vocal processing including pitch correction, isolation, noise reduction, and harmony generation
deep-learning
pluginagentmarketplace
Neural networks, CNNs, RNNs, Transformers with TensorFlow and PyTorch. Use for image classification, NLP, sequence modeling, or complex pattern recognition.
modal
eyadsibai
Use when "Modal", "serverless GPU", "cloud GPU", "deploy ML model", or asking about "serverless containers", "GPU compute", "batch processing", "scheduled jobs", "autoscaling ML"
gemini-api
diskd-ai
Google Gemini API integration for building AI-powered applications. Use when working with Google's Gemini API, Python SDK (google-genai), TypeScript SDK (@google/genai), multimodal inputs (image, video, audio, PDF), thinking/reasoning features, streaming responses, structured outputs with JSON schemas, multi-turn chat, system instructions, image generation (Nano Banana), video generation (Veo), music generation (Lyria), embeddings, document/PDF processing, or any Gemini API integration task. Triggers on mentions of Gemini, Gemini 3, Gemini 2.5, Google AI, Nano Banana, Veo, Lyria, google-genai, or @google/genai SDK usage.
ai-model-orchestrator
famaoai-creator
Dynamically selects the optimal AI model based on task complexity, cost, and latency. Routes requests to Gemini, GPT-4, Claude, or local LLMs to maximize efficiency.
omp-surface-choice
srobroek
Use when deciding whether guidance becomes a TTSR rule, a tool_call gate, a registered tool, a slash command, a skill, or stays prose.
simpo-training
magnetoid
Simple Preference Optimization for LLM alignment. Reference-free alternative to DPO with better performance (+6.4 points on AlpacaEval 2.0). No reference model needed, more efficient than DPO. Use for preference alignment when want simpler, faster training than DPO/PPO.