output-styles
Understanding output styles in Claude Code (deprecated feature). Use when user asks about customizing Claude's behavior, output styles, or explanatory/learning modes.
Understanding output styles in Claude Code (deprecated feature). Use when user asks about customizing Claude's behavior, output styles, or explanatory/learning modes.
Apple Foundation Models framework for on-device AI, @Generable macro, guided generation, tool calling, and streaming. Use when user asks about on-device AI, Apple Intelligence, Foundation Models, @Generable, LLM, or local machine learning.
Build cost-optimized chat applications with Azure OpenAI Model Router that intelligently routes queries to the most appropriate model based on complexity, achieving up to 60% cost savings.
Deep research with sequential thinking and parallel agents
Set up Ralph for autonomous feature development. Use when starting a new feature that Ralph will implement. Triggers on: ralph, set up ralph, start ralph, new ralph feature, ralph setup. Chats through the feature idea, creates tasks with dependencies, and sets up everything for Ralph to run.
Check status of running subagents including token usage and context limits
Generate culturally diverse names for examples, mock data, and test fixtures. Includes edge-case names that catch bugs.
Use when writing or refactoring Python code for LangGraph v1.x + LangChain v1: Graph API + Functional API import paths, checkpoint/store usage, and migrations off deprecated prebuilts (create_react_agent -> langchain.agents.create_agent).
TOON Format Specialist - YAML-based token-efficient agent/workflow definitions inspired by BMAD Method patterns
Create new Agent Skills interactively or from templates. Use when user wants to create, generate, scaffold, or build a new skill, or mentions creating skills, writing skills, skill templates, skill development.
Helper agent to scaffold new Gemini CLI skills. Use this when the user wants to "make a new skill", "add a skill", or "teach you a new capability" via the skill system.
Multi-Model Orchestration - Guide for orchestrating multi-model agents
Apply modern AI/LLM development best practices: staying current on models, prompt/context engineering, architecture patterns, stack decisions, evaluation, and production deployment. Use when building AI features, selecting models, writing prompts, reviewing LLM code, or discussing AI architecture.
Amazon Bedrock Runtime API for model inference including Claude, Nova, Titan, and third-party models. Covers invoke-model, converse API, streaming responses, token counting, async invocation, and guardrails. Use when invoking foundation models, building conversational AI, streaming model responses, optimizing token usage, or implementing runtime guardrails.
Templates and patterns for common ML training scenarios including text classification, text generation, fine-tuning, and PEFT/LoRA. Provides ready-to-use training configurations, dataset preparation scripts, and complete training pipelines. Use when building ML training pipelines, fine-tuning models, implementing classification or generation tasks, setting up PEFT/LoRA training, or when user mentions model training, fine-tuning, classification, generation, or parameter-efficient tuning.
Deploying fine-tuned models for production inference using native kernel optimization, vLLM, or SGLang. Triggers: inference, serving, vllm, sglang, for_inference, model merging, openai api.
Expert in managing the "Memory" of AI systems. Specializes in Vector Databases (RAG), Short/Long-term memory architectures, and Context Window optimization. Use when designing AI memory systems, optimizing context usage, or implementing conversation history management.
Fine-tune LLMs using the Tinker API. Covers supervised fine-tuning, reinforcement learning, LoRA training, vision-language models, and both high-level Cookbook patterns and low-level API usage.
Entry point for Codex-discoverable skills used by the Run-Smart AI coach.