Use when wanting to interact with any LLM - Explains available inference endpoints so the agent selects suitable models.
Use when discovering, downloading, quantizing, serving, benchmarking, or operating LLMs locally or on GPU infrastructure with Hugging Face, llama.cpp, vLLM, W&B, or lm-eval.
Use this when: run a model locally, which model fits my GPU, my inference is too slow, serve LLM in production, how to quantize a model, too many concurrent users, self-host an…
LLM.int8() and bitsandbytes 4-bit quantization (NF4/FP4) for memory-efficient LLM loading. Mixed-precision decomposition, QLoRA fine-tuning, double quantization, and GPU OOM…
Guide for using LLM utilities in speedy_utils, including memoized OpenAI clients and chat format transformations.
Intégration de LLMs dans des applications via API. Se déclenche avec "API OpenAI", "Claude API", "intégrer un LLM", "GPT dans mon app", "Ollama", "LLM local", "streaming" — from…
Advanced LLM jailbreaking techniques, safety mechanism bypass strategies, and constraint circumvention methods
Document both INPUT and OUTPUT JSON schemas for LLM-driven features to prevent data flow confusion
Use when comparing two or more code implementations against a spec or requirements doc. Triggers on \"which repo is better\", \"compare these implementations\", \"evaluate both…
Comprehensive guide to using LLMs as judges for automated evaluation including prompt patterns, calibration, bias reduction, and multi-judge ensembles
分析 llm-gateway 代理产生的请求/响应日志。当用户说"分析日志"、"查看 LLM 请求"、"对比 session"、"检查 token 用量"、"日志里有什么"、"帮我看看 data 目录"、"哪个请求失败了"、"找一下 session 的请求"等涉及 LLM 网关日志分析的场景时使用此 skill。即使用户只是笼统地说"看看日志"或"data…
Use when the user runs /llm-login or wants to authenticate an LLM provider (GitHub Copilot, Kimi, MiniMax, GLM, Qwen), check auth status, or logout from a provider.
Dynamic LLM model management. Switch between model tiers (flagship, worker, efficiency, reasoning, coding) at runtime to optimize for intelligence, speed, or cost.
Access Claude, Gemini, Kimi, GLM and 100+ LLMs via inference.sh CLI using OpenRouter. Models: Claude Opus 4.5, Claude Sonnet 4.5, Claude Haiku 4.5, Gemini 3 Pro, Kimi K2, GLM-4.6,…
Building AI-powered NPCs that maintain personality, remember conversations, and never break characterUse when "npc dialogue, ai npc, llm npc, character dialogue, npc personality,…
Classify whether user intent was satisfied in a Datadog LLM Obs trace or session. Three modes: (1) session_id — classify a single CMD+I assistant session with RUM; (2) trace_id —…
LLM Operations -- RAG, embeddings, vector databases, fine-tuning, prompt engineering avancado, custos de LLM, evals de qualidade e arquiteturas de IA para producao. — from renat
LLM Operations -- RAG, embeddings, vector databases, fine-tuning, prompt engineering avancado, custos de LLM, evals de qualidade e arquiteturas de IA para producao.
Optimize websites for AI assistant recommendations. ChatGPT, Gemini, Perplexity, Claude. Get cited in AI answers.
Use when creating content that must be discoverable by AI search engines (ChatGPT, Perplexity, Gemini).
Assessing privacy risks in large language model outputs including training data memorisation, PII leakage in generated text, prompt injection leading to data extraction, and…
Review or plan changes that affect prompts, model output validation, fallback behavior, or LLM-driven decisions.
Inspect LLM torch profiler traces at forward-pass, layer, and kernel level. Use when you need layer timings, anchor-kernel boundaries, representative kernel flows, or Perfetto…
ドキュメント・プロンプト・スキル定義が、Claude 以外のLLM（GPT-4o / Gemini / Codex 等）でも意図通りに動作するかを評価する。Claude 固有の解釈・暗黙前提・スタイル依存を検出し、移植耐性を高める改善案を提示する。他LLMへ移植する前に使う。
Write social media posts for X (Twitter) and LinkedIn that avoid AI tells, pass platform algorithms, and drive engagement.
Plan and execute large language model pretraining from data preparation to checkpoint management
Guide and workflow for LLM Prompt Engineering - System Instruction Best Practices. Use when you need LLM Prompt Engineering - System Instruction Best Practices.
Use when improving prompts for any LLM. Applies proven prompt engineering techniques to boost output quality, reduce hallucinations, and cut token usage.
Use when improving prompts for any LLM. Applies proven prompt engineering techniques to boost output quality, reduce hallucinations, and cut token usage.
面向个人开发者的OpenAI API轻量调用工具。核心能力: - 对话补全(chat completions)与文本生成 - 文件上传与基础向量存储管理 - DALL-E 图像生成与编辑 - 助手(Assistants)基础创建与对话 - 文本嵌入(embeddings)生成
面向团队与企业用户的 llm-provider API 全功能管理工具。核心能力: - 涵盖免费版全部能力(对话补全、图像生成、助手管理) - 批量任务(Batch API)大规模异步处理 - 模型微调(Fine-tuning)定制化训练 - 评估(Evaluations)质量度量与回归测试 - 向量存储(Vector Stores)高级 — from…
Before stating any specific fact about an LLM provider's product — models, pricing, limits, context windows, SDK/API — for OpenAI, Gemini, Claude & others, verify against official…
Set up a new LLM provider connection, discover valid model ids, create and validate a model profile — securely and end-to-end via CLI.
基于 Whisper v1 稳定版本的本地语音转文字工具(免费版)。核心能力: - v1 稳定版 CLI 转录能力,接口简洁可靠 - 支持 mp3 / m4a / wav 等常见音频格式 - 输出 txt / srt / vtt 等多种字幕格式 - 内置翻译模式(任意语言转英文) - 模型自动下载与本地缓存
Use for LLM quantization — MLX mixed precision, GPTQ/AWQ, sensitivity, memory, perplexity.
Comprehensive LLM model evaluation and ranking system. Use when users ask to compare language models, find the best model for a specific task, understand model capabilities, get…
Restructure Web-UI / human-triggered tasks into CLI + file-output loops the LLM can iterate alone. Open LLM-side observability — structured logs, file dumps, addressable…
通过 LaTeX 生成面向客户交付的「企业大模型技术方案」类 PDF（学术/公文风格排版），包含能力与客户系统映射、集成架构示意、分场景收益与 KPI、部署/安全/运维与版本控制表。适用于投标/POC 技术附件、客户技术建议书、需版式统一且可追溯的交付文档场景。
LLM streaming response patterns. Use when implementing real-time token streaming, Server-Sent Events for AI responses, or streaming with tool calls.
Build production LLM streaming UIs with Server-Sent Events, real-time token display, cancellation, error recovery. Handles OpenAI/Anthropic/Claude streaming APIs.
Get reliable JSON, enums, and typed objects from LLMs using response_format, tool_use, and schema-constrained decoding across OpenAI, Anthropic, and Google APIs.
Graceful rate limit handling with Ollama fallback. Notifies on rate limits, offers local model switch with confirmation for code tasks.
Summarize and navigate this repo with the llm-tldr CLI (tree/structure/context/semantic search). Use when you need LLM-ready summaries of files or symbols, fast codebase…
See the main LLM Cost Optimization skill for comprehensive coverage of token economics and optimization strategies.
Evidence-based LLM parameter configurations for theorem proving, code generation, and reasoning tasks. Use when tuning temperature, top_p, or max_tokens.
Design UI for LLM-powered features — streaming response display, hallucination states, RAG source attribution, generative UI patterns, and AI trust calibration.
verificação independente e adversarial de uma task DEPOIS que o DoD está todo verde, antes de declarar "feito"; ativa ao fechar qualquer task técnica, ao responder "deu ok?" /…
Karpathy's LLM Wiki — build and maintain a persistent, interlinked markdown knowledge base. Ingest sources, query compiled knowledge, and lint for consistency.
自動化 LLM Wiki 知識編譯技能。當啟動時，它會掃描指定目錄下的新檔案，並依照 Karpathy 的 LLM Wiki 邏輯（Ingest -> Summarize -> Compile -> Log）將其整合進 Obsidian 知識庫。
Maintains docs/ in the Karpathy-style LLM-wiki standard. Generates docs/llms.txt (flat machine-readable index conforming to llmstxt.org), validates broken links, enforces a…
Register a project's wiki/ directory into the cross-project aggregate via symlink (whitelist mode). Use when user invokes /llm-wiki-link or asks to add a project to the…
Operate and maintain the LLM Wiki knowledge system in the Megingjord source workspace.
Operate an LLM Wiki knowledge system in any workspace using the Karpathy pattern
个人知识库构建系统（基于 Karpathy llm-wiki 方法论）。让 AI 持续构建和维护你的知识库， 支持多种素材源（网页、推特、公众号、小红书、知乎、YouTube、PDF、本地文件）， 自动整理为结构化的 wiki。 触发条件：用户明确提到"知识库"、"wiki"、"llm-wiki"，或要求对已初始化的知识库执行…
Use when producing written artifacts for humans — catches default LLM writing patterns.
LLM4S Scala functional LLM interfaces with Effect system integration. Use when building LLM applications in Scala with ZIO or Cats Effect, implementing type-safe AI pipelines with…
Route any prompt to the best-performing LLM using peer-reviewed council rankings from LLM Council
Default operating skill for llmdoc-enabled projects. Use when a project has llmdoc/, when initializing llmdoc, when updating project knowledge, or when Codex CLI hooks should…
Codex-native entry skill for bootstrapping llmdoc. Use this when you want the /llmdoc:init workflow in Codex.
Detect local hardware (RAM, CPU, GPU/VRAM) and recommend the best-fit local LLM models with optimal quantization, speed estimates, and fit scoring.