LLM

Control omi reasoning effort level to balance depth, rigor, latency, and token cost.

Select the correct Ollama base model (and adapters) based on task type, resource fit, and registry availability. Use to translate Modelfile FROM/ADAPTER decisions into agent behavior.

Prompt debugging skills for LLM developers. Convert conversations from OpenAI, Anthropic, Google, and Helicone into .echo format, and integrate .echo export into your apps. Use with EchoSpace — the local-first prompt debugging workspace.

Finetune LLMs to speak Slipstream natively - complete guide with GLM-4-9B

**프롬프트 엔지니어링 마스터**: AI 모델(Claude, ChatGPT 등)에게 최적의 결과를 이끌어내는 프롬프트 설계 스킬. 역할지정, 체인오브소트(CoT), 퓨샷, 제로샷, 메타프롬프트 등 26가지 기법을 실무 맥락에 맞게 적용합니다. - MANDATORY TRIGGERS: 프롬프트, prompt, 프롬프트 엔지니어링, AI 질문법, AI 활용법, 프롬프팅, 역할 부여, CoT, chain of thought

Expert guidance for Cerebras Inference, the ultra-fast LLM inference service powered by the world's largest chip (Wafer-Scale Engine). Helps developers integrate Cerebras' API for applications requiring the fastest possible token generation — real-time chat, code completion, and interactive AI experiences.

Master advanced prompt engineering techniques to maximize LLM performance, reliability, and controllability in production. Use when optimizing prompts, improving LLM outputs, or designing productio...

Research latest LLM benchmarks and update primary/fallback provider configurations if better models are available.

Expert guidance for GRPO/RL fine-tuning with TRL for reasoning and task-specific model training

Use when implementing function calling, tool use, or agents with LLMs - unified tool API works across OpenAI, Anthropic, Google, and Ollama with consistent tool definition and execution patterns

Build with Claude Messages API using structured outputs (v0.69.0+, Nov 2025) for guaranteed JSON schema validation. Covers prompt caching (90% savings), streaming SSE, tool use, model deprecations (3.5/3.7 retired Oct 2025). Use when: building chatbots/agents with validated JSON responses, or troubleshooting rate_limit_error, structured output validation, prompt caching not activating, streaming SSE parsing.

Use this skill when you are writing commands, hooks, skills for Agent, or prompts for sub agents or any other LLM interaction, including optimizing prompts, improving LLM outputs, or designing production prompt templates.

Gemma Domain Trainer (Prototype)

LLMs, prompt engineering, RAG systems, LangChain, and AI application development

Use this skill when you writing commands, hooks, skills for Agent, or prompts for sub agents or any other LLM interaction, including optimizing prompts, improving LLM outputs, or designing production prompt templates.

Recursive Meta-Prompting (RMP) implementation with unified categorical syntax. Supports @mode:iterative, @quality: thresholds, >=> Kleisli composition, and comonadic context extraction. Use when implementing iterative prompt improvement, quality-gated generation loops, or applying categorical fixed-point semantics with convergence guarantees.

Extract structured data from LLM responses with Pydantic validation, retry failed extractions automatically, parse complex JSON with type safety, and stream partial results with Instructor - battle-tested structured output library

Structured prompt construction and quality validation for Jerry Framework. Invoke when building structured prompts, generating NPT-009/NPT-013 constraints, or scoring prompt quality. Guides users through the 5-element prompt anatomy, generates formatted constraints with XML wrapping, and scores prompts against the 7-criterion rubric.

Track popular/new open-source LLMs and update docs/model_coverage.mdx with their kernel support status. Use when discovering new models to add to the coverage tracker, checking if a specific model is covered, or refreshing model coverage documentation.

Advanced Hive Mind collective intelligence system for queen-led multi-agent coordination with consensus mechanisms and persistent memory

LangSmith 통합 및 LLM Observability 가이드. 토큰 추적, 비용 계산, Run Tree, Tracing 데코레이터, OTEL 연동 구현 시 참조. "langsmith", "tracing", "observability", "token usage", "cost tracking" 키워드로 트리거.

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.

Provides tool and function calling patterns with LangChain4j. Handles defining tools, function calls, and LLM agent integration. Use when building agentic applications that interact with tools.

Deploys ML and LLM models on TrueFoundry with GPU inference servers (vLLM, TGI, NVIDIA NIM). Uses YAML manifests with `tfy apply`. Use when serving language models, deploying Hugging Face models, or hosting GPU-accelerated inference endpoints.

Sub-skill técnica de Yann LeCun. Cobre CNNs, LeNet, backpropagation, JEPA (I-JEPA, V-JEPA, MC-JEPA), AMI (Advanced Machinery of Intelligence), Self-Supervised Learning (SimCLR, MAE, BYOL),...

指导在 CodeSpirit 项目中集成 AI 功能的完整开发流程。包括 AI 表单填充、AI 长任务处理、LLM 集成和提示词工程。当用户需要添加 AI 功能、集成 LLM、或开发 AI 驱动的业务功能时使用。

Operational prompt engineering for production LLM apps: structured outputs (JSON/schema), deterministic extractors, RAG grounding/citations, tool/agent workflows, prompt safety (injection/exfiltration), and prompt evaluation/regression testing. Use when designing, debugging, or standardizing prompts for Codex CLI, Claude Code, and OpenAI/Anthropic/Gemini APIs.

Extend context windows of transformer models using RoPE, YaRN, ALiBi, and position interpolation techniques. Use when processing long documents (32k-128k+ tokens), extending pre-trained models beyond original context limits, or implementing efficient positional encodings. Covers rotary embeddings, attention biases, interpolation methods, and extrapolation strategies for LLMs.

Remove refusal behaviors from open-weight LLMs using OBLITERATUS — mechanistic interpretability techniques (diff-in-means, SVD, whitened SVD, LEACE, SAE decomposition, etc.) to excise guardrails while preserving reasoning. 9 CLI methods, 28 analysis modules, 116 model presets across 5 compute tiers, tournament evaluation, and telemetry-driven recommendations. Use when a user wants to uncensor, abliterate, or remove refusal from an LLM.

Open-source AI observability platform for tracing, evaluating, and improving LLM applications with OpenTelemetry integration

World-class expert in prompt engineering, LLM fine-tuning, RAG systems, and AI/ML workflows. Use when crafting prompts, designing AI agents, building knowledge bases, implementing retrieval systems, or optimizing LLM performance at production scale.

Evaluate LLM systems using automated metrics, LLM-as-judge, and benchmarks. Use when testing prompt quality, validating RAG pipelines, measuring safety (hallucinations, bias), or comparing models for production deployment.

Adaptive multi-model AI roundtable. Runs up to 4 AI models (configurable) in 2 debate rounds with cross-critique and formal consensus scoring. Requires a configured Anthropic provider (Claude Opus recommended). Optionally adds GPT-5.3 Codex (OpenAI), Grok 4, and Gemini 3.1 Pro via Blockrun proxy. Works with Claude-only fallback if optional providers are unavailable. Writes results to local filesystem. Debate panel agents are persistent thread sessions; meta-panel and synthesis agents are one-shot.

Build apps with the Claude API or Anthropic SDK. TRIGGER when: code imports `anthropic`/`@anthropic-ai/sdk`/`claude_agent_sdk`, or user asks to use Claude API, Anthropic SDKs, or Agent SDK. DO NOT TRIGGER when: code imports `openai`/other AI SDK, general programming, or ML/data-science tasks.

Teach Mandarin using an LLM-first pedagogy, backed by a ZFC/Unix-style local engine (`xuezh`) that stores facts, runs mechanical transforms, and produces bounded reports/audio artifacts. Use for review, speaking/tones, graded input, and HSK audits.

Google Gemini SDK (google-genai) 활용 가이드. Gemini 3/2 모델, Structured Output, Function Calling, Google Search Grounding, 이미지 생성 구현 시 참조. "gemini", "google ai", "genai", "google search", "imagen" 키워드로 트리거.

Design LLM applications using LangChain 1.x and LangGraph for agents, memory, and tool integration. Use when building LangChain applications, implementing AI agents, or creating complex LLM workflows.

Run LLMs on Apple Silicon with MLX/mlx_lm - unified memory, 4-bit quantization, streaming generation, prompt caching. Optimal for M-series chips.

Automated hypothesis generation and testing using large language models. Use this skill when generating scientific hypotheses from datasets, combining literature insights with empirical data, testing hypotheses against observational data, or conducting systematic hypothesis exploration for research discovery in domains like deception detection, AI content detection, mental health analysis, or other empirical research tasks.

Complete API for Google NotebookLM - full programmatic access including features not in the web UI. Create notebooks, add sources, generate all artifact types, download in multiple formats. Activates on explicit /notebooklm or intent like "create a podcast about X"

ElizaOS - TypeScript framework for building autonomous AI agents with multi-platform support (Discord, Telegram, Twitter, Farcaster), blockchain integration (EVM, Solana), plugin architecture, multi-agent orchestration, and 90+ community plugins

This skill should be used when users want to train or fine-tune language models using TRL (Transformer Reinforcement Learning) on Hugging Face Jobs infrastructure. Covers SFT, DPO, GRPO and reward modeling training methods, plus GGUF conversion for local deployment. Includes guidance on the TRL Jobs package, UV scripts with PEP 723 format, dataset preparation and validation, hardware selection, cost estimation, Trackio monitoring, Hub authentication, and model persistence. Should be invoked for tasks involving cloud GPU training, GGUF conversion, or when users mention training on Hugging Face Jobs without local GPU setup.

Deep expertise in Azure OpenAI Service — deploy and manage GPT-4o, GPT-4, GPT-3.5-Turbo, Embeddings, DALL-E, Whisper, and TTS models. Covers Standard, Provisioned-Managed, and Global Standard deployment types, fine-tuning workflows, content filtering policies, prompt engineering patterns, Batch API, quota management, and secure production architectures. Uses az cognitiveservices CLI and Azure OpenAI REST API for all operations.

LLM skill for parsing Dockerfile and Makefile into NonDeterministicValues YAML format

Professional market data and AI APIs via x402 micropayments — no API key, no signup, no subscription. Pay per call with USDC on Base. 234+ endpoints across 16 provider groups: Web3 intelligence (RootData — 70,000+ crypto projects, VCs, funding rounds, trending, KOLs, personnel movement), crypto market data (CoinAnk derivatives analytics, nofxos.ai AI signals, CoinMarketCap quotes/listings/DEX/MCP), US stocks & options (Alpaca, Polygon, Alpha Vantage), China A-shares (Tushare), forex & global time-series (Twelve Data), and AI inference (GPT-5.4/5.3, Claude Opus/Sonnet/Haiku, DeepSeek V3/Reasoner, Qwen3-Max/Plus/Turbo/Flash/Coder/VL, Gemini 3.1 Pro/Flash/2.5, Grok-4.1/4/3-Mini, Kimi K2.5/K2, embeddings, DALL-E). One wallet, instant access to any paid API — no registration ever required.

Agente que simula Yann LeCun — inventor das Convolutional Neural Networks, Chief AI Scientist da Meta, Prêmio Turing 2018. Use quando quiser: perspectivas sobre deep learning e visão...

Tari Ootle development instructions for Google Gemini

Agente que simula Andrej Karpathy — ex-Director of AI da Tesla, co-fundador da OpenAI, fundador da Eureka Labs, e o maior educador de deep learning do mundo. Use quando quiser: aprender deep...

Tari Ootle development instructions for OpenAI Codex

Use when implementing on-device AI with Apple's Foundation Models framework — prevents context overflow, blocking UI, wrong model use cases, and manual JSON parsing when @Generable should be used. iOS 26+, macOS 26+, iPadOS 26+, axiom-visionOS 26+