В каталоге 185 367 навыков в 14 категориях.
This skill should be used when building production LLM applications in any language. It applies when implementing predic...
Use when building LLM applications: prompt engineering, structured output, agents, RAG integration, memory management, o...
Expert LLM architect specializing in large language model architecture, deployment, and optimization. Masters LLM system...
Use when user needs LLM system architecture, model deployment, optimization strategies, and production serving infrastru...
Orchestrate conversations between two LLMs via OpenRouter and compare their conversational personalities, behavioral pat...
Detects common LLM coding agent artifacts in codebases. Identifies test quality issues, dead code, over-abstraction, and...
Execute programs on a compiled transformer stack machine where every instruction fetch and memory read is a parabolic at...
Design pattern for LLM-as-judge evaluators: binary checks as evidence, one named verdict.
Force-clamp free-form LLM output to binary yes/no with keyword matching and a fallback default for constrained environme...
Extract final numeric answers from LaTeX \boxed{} notation in LLM math reasoning output, scanning matches in reverse for...
Implement multi-layer LLM caching with exact match, semantic similarity, and provider-side prompt caching. Reduce API co...
Multi-level caching strategies for LLM applications - semantic caching (Redis), prompt caching (Claude/OpenAI native), c...
Analyze LLM token logprobs and calibration. Use for per-decision confidence, ECE, Brier scores, reliability diagrams, an...
External LLM invocation. Triggered ONLY by @council,@probe,@crossref,@gpt,@gemini,@grok,@qwen.
Find LLM API calls that lack observability - no logging, error handling, timeout, or cost tracking. Trigger: /llm-obs
Instrument LLM API calls with proper spans, tokens, and latency
Gate resumption of any multi-day LLM batch-scoring campaign that calls an unpinned model alias (deepseek-chat, gpt-*-lat...
Plan and record evidence-backed LLM and AI-search citation checks across approved assistant/search surfaces with timesta...
LLM-based zero-shot and few-shot classification for flexible intent detection
Centralized AI-readable documentation repository with 245+ frameworks and tools. Use to find documentation, add new sour...
Behavioral guardrails to reduce common LLM coding mistakes with a caution-first approach.
Write effective LLM prompts, commands, and agent instructions. Goal-oriented over step-prescriptive. Role + Objective +...
Configure RuVLLM local inference with model selection, MicroLoRA fine-tuning, and SONA adaptation
USE FOR RAG/LLM grounding. Returns pre-extracted web content (text, tables, code) optimized for LLMs. GET + POST. Adjust...
Guidelines for working with LLM context stored in the .llm/ directory.
LLM cost controller: audits all LLM API calls in a codebase, detects cost anti-patterns (expensive models, unlimited tok...
Model the cost and latency of an LLM feature before it ships and surprises the bill. Use when asked to estimate LLM API...
Strategies for managing and reducing costs in LLM-powered applications, from token economics to RAG architectures.
Reduce LLM API and infrastructure costs through model selection, prompt caching, batching, caching, quantization, and se...
Reduce LLM API costs without sacrificing quality. Covers prompt caching (Anthropic), local response caching, prompt comp...
Use when an LLM feature costs too much. Covers prompt caching, context reduction, model routing, batching, output limits...
Use proactively whenever LLM API costs come up -- or should. Triggers include: 'my AI costs are too high', 'optimize tok...
Analyze and reduce LLM spend: read usage breakdowns by call site, model, and inference profile, understand single-winner...
Orchestrate a configurable, multi-member CLI planning council (Codex, Claude Code, Gemini, OpenCode, or custom) to produ...
[Decision Support] Use when pressure-testing an irreversible, high-stakes decision with adversarial AI advisors.
Multi-LLM collaborative brainstorming and planning. Use when user explicitly requests consultation with multiple AI mode...
Convene a cross-vendor LLM council — the session is Chairman and spawns subagents each pinned to a DIFFERENT model famil...
Convene an internal "council" of distinct senior-engineer personas to analyze a problem, stress-test an approach, and pr...
Orchestrate multi-LLM parallel debate and synthesis. Dispatches prompts to available omega CLI wrappers in parallel, col...
Provider-agnostic multi-LLM deliberation. Three phases — independent responses, cross-model anonymized ranking, chairman...
Run Fireworks-hosted open-weight model councils that compare responses and synthesize a final answer.
Run any question, idea, or decision through a council of 5 AI advisors who independently analyze it, peer-review each ot...
This skill should be used when Claude needs to "formulate a request for external feedback", "craft a question for ChatGP...
Check whether a website's robots.txt allows the AI crawlers that decide visibility in ChatGPT Search, Perplexity, Claude...
Diagnoses LLM output failures including hallucinations, constraint violations, format errors, and reasoning issues. Prov...
Use when lLM deployment and serving — vLLM, Ollama, TGI, llama.cpp. Model quantization, GPU optimization, API serving. U...
Generate the LLM-friendly API cheat sheets in docs_llm/api from Doxygen XML. Use when public headers or their Doxygen co...
Write token-efficient documentation for LLM context. Use when creating CLAUDE.md, README, technical docs, agent instruct...
Fetch LLM-optimized documentation from llms.txt endpoints for up-to-date API references
Extract structured data from construction documents using LLMs. Process RFIs, submittals, contracts, specifications. Con...
Review AI systems for LLM-specific denial-of-service vulnerabilities including token flooding, recursive prompt expansio...
Patterns for building LLM applications - prompt engineering, RAG pipelines, cost optimization, multi-model routing, and...
Measure an LLM feature instead of vibe-checking it. Five modes. ANALYZE: read 20–30 real outputs, write one line each, g...
LLM生成システムの検証設計スキル。assay-kitフレームワークを活用し、LLM特有の失敗モード(幻覚、例への過学習、部分的処理)を考慮した総合的なテストケース設計を支援する。使用タイミング:- LLMベースのワークフロー/エージェント...
Pick and wire the right LLM evaluation tool for a repository - Promptfoo (prompt regression and CI gating on local llama...
Implement comprehensive evaluation strategies for LLM applications using automated metrics, human feedback, and benchmar...
Use when measuring LLM application performance systematically; Comparing different models or prompts
Design and run an evaluation for an AI or agent feature using representative tasks, explicit success criteria, failure t...
LLM testing and evaluation tools including promptfoo, trulens, and evals frameworks
Design, run and interpret evaluations of LLM-based systems - build a task-specific golden set, pick metrics (exact match...
use when evaluating LLMs, prompts or workflows
Use when measuring the quality of an LLM feature. Covers building an evaluation set, choosing metrics, LLM-as-judge and...
Master comprehensive evaluation strategies for LLM applications, from automated metrics to human evaluation and A/B test...
Use this skill when you need to design LLM evaluation datasets, judges, metrics, and human-review boundaries; triggers i...
LLM 中枢服务层设计规范。修改 services/llm/ 目录下任何文件时必须加载。涉及 LLM Provider 适配、工厂模式、API 调用方式变更时触发。
LLM Fine-Tuning expert. Covers LoRA, QLoRA, PEFT, dataset preparation, Hugging Face Trainer/TRL, RLHF, DPO, quantization...
Set up infrastructure for fine-tuning LLMs with QLoRA, LoRA, and full fine-tuning using Hugging Face TRL, Axolotl, and d...
Fine-tune large language models efficiently using LoRA, QLoRA, and PEFT methods. Use for domain adaptation, instruction...
Production-grade skill package for Llm Fine Tuning Lora in data-mlops.
Expert guide for AI FinOps, dynamic model routing (LiteLLM, Portkey), cost optimization, and latency-based fallback arch...
Clarifies inputs, outputs, success criteria, decisions, and unresolved conditions so downstream agents can execute witho...
Clarifies inputs, outputs, success criteria, decisions, and unresolved conditions so downstream consumers can execute wi...
Design human-visible, machine-readable website layers for AI assistants, RAG, and agentic search. Use this skill when th...
LLM function calling (also known as tool use) enables Large Language Models to interact with external systems by calling...
LLM-powered quality verification using prompt hooks. Validates commit messages, code patterns, and conventions using AI...
LLM gateway and routing configuration using OpenRouter and LiteLLM. Invoke when: - Setting up multi-model access (OpenRo...
Diagnose whether an LLM classifier's validation-gate failure is GOLD-BOUND before spending on prompt revision or model c...
LLM content governance and compliance standards. Use when llm governance guidance is required.
Comprehensive guide for LLM safety and guardrails implementation.
Specify the safety and reliability guardrails for an LLM feature before it ships. Use when asked to define LLM guardrail...
Finding and accessing AI/LLM model brand icons from lobe-icons library. Use when users need icon URLs, want to download...
Use when wanting to interact with any LLM - Explains available inference endpoints so the agent selects suitable models.
Guidance for optimizing LLM inference request batching and scheduling problems. This skill applies when designing batch...
Auto-scale LLM inference clusters on Kubernetes using KEDA, custom GPU metrics, and horizontal pod autoscaling. Handle t...
Guide for using LLM utilities in speedy_utils, including memoized OpenAI clients and chat format transformations.
LLM integration patterns for Claude, GPT, Gemini, and Ollama. Activate for AI API integration, prompt engineering, token...
Use when integrating an LLM API into an application. Covers streaming, retries and rate limits, timeouts, caching, fallb...
from langchainopenai import ChatOpenAI
Advanced LLM jailbreaking techniques, safety mechanism bypass strategies, and constraint circumvention methods. Use for...
LLM-as-judge methodology for comparing code implementations across repositories. Scores implementations on functionality...
Calibrate LLM judges: rubrics, calibration sets, inter-annotator agreement, drift detection and judge/teacher separation...
Shares KV cache from a common prefix (context+question) across multiple answer suffixes for efficient multi-choice scori...
Lokales LLM in zid: KI-Chat über llama-server, Selbsteinrichtung von Engine und Modell, Engine- und Modellwahl, gepinnte...
Comprehensive guide for deploying LLMs locally using Ollama, vLLM, and llama.cpp. Local deployment offers privacy, cost...
Claude acts as manager/architect while delegating all coding to external LLM CLIs (Gemini, Codex, Qwen). Claude never wr...
Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services...
当使用者说"让 AI 写谱不要直接生成音频"、"做一个 AI 作曲的后端"、"LLM 输出 MIDI 怎么做"、"生成的音符老是越界/被截断"、"要能导出 .mid 给 DAW"时使用。用 LLM(BYOK / OpenAI 兼容接口)从零...
Choosing the right model for the task — power vs. cost vs. speed.
Access Claude, Gemini, Kimi, GLM and 100+ LLMs via inference.sh CLI using OpenRouter. Models: Claude Opus 4.5, Claude So...