Use when designing or reviewing list views, compact layouts, highlights panels, and search-result presentation so users can scan, find, and act on records quickly across desktop…
Subscribe to and listen for Tauri v2 events in the frontend via @tauri-apps/api/event, including typed handlers and cleanup.
Add the Lite-Basic (Tier 2) sprite atlas to an EXISTING Codogotchi pet that already has a Codex `spritesheet.webp`.
Testing patterns for litefs-py and litefs-django. Use when writing tests, setting up fixtures, understanding test organization, or configuring pytest marks.
Unified LLM API with LiteLLM. Call 100+ LLM providers with one interface. Use for multi-provider AI, cost optimization, fallbacks, and LLM gateway deployment.
Administer a LiteLLM proxy via its management REST API (keys, teams, users, orgs, models, budgets, spend) — a 447-path control plane where ~63% of paths are undocumented, 91 are…
When calling LLM APIs from Python code. When connecting to llamafile or local LLM servers. When switching between OpenAI/Anthropic/local providers.
Paperclip-style multi-source literature search across arXiv, Crossref, local JSON corpora, and (opt-in) the Paperclip API.
Auto-activate for guards=, Guard, ASGIConnection, JWTAuth, JWTCookieAuth, SessionAuth, role or tenant checks, or WebSocket auth. Not for frontend route protection.
Implements and trains LLMs using Lightning AI's LitGPT with 20+ pretrained architectures (Llama, Gemma, Phi, Qwen, Mistral).
Receive and verify Lithic webhooks. Use when setting up Lithic event subscriptions, debugging Standard Webhooks signature verification, or handling card and money-movement events…
Run a senior-level Java/Spring backend live-coding interview practice, guiding the user with Socratic prompts, strict code review (readability, edge cases, data structure — from…
Symfony UX LiveComponent for dynamic server-rendered UI. Use when building interactive components that re-render via AJAX, real-time forms, data binding, live validation, or…
Verify live ThumbGate.app service health, trace header propagation, and production deployment status using curl + BrowserOS.
Delivering real-time updates to users via WebSocket, SSE, or Push API for live notification systems with proper architecture, queuing, and delivery mechanisms.
Use when building real-service end-to-end tests with fixtures, cleanup, rate limits, and evidence. Triggers:
Génère un site communautaire en temps réel à partir des contributions du chat Twitch. Chaque contribution = code généré + git commit + push + auto-deploy Cloudflare Pages.
Real-time task progress tracking with live push notifications. MANDATORY for ALL multi-step tasks (>30s or >2 tool calls).
Comprehensive guide for building functional tools for LiveKit voice agents using the @function_tool decorator.
Build and review production-grade web and mobile frontends using LiveKit with Next.js. Covers real-time video/audio/data communication, WebRTC connections, track management, and…
LiveKit is an open-source, scalable WebRTC-based real-time communication server written in Go. It provides multi-user conferencing, streaming, and data channels with client SDKs…
Build voice AI agents with LiveKit Agents SDK. Use when the user asks to "build a voice agent", "create a LiveKit agent", "add voice AI", "implement handoffs", "structure agent…
Reviews Phoenix LiveView code for lifecycle patterns, assigns/streams usage, components, and security. Use when reviewing LiveView modules, .heex templates, or LiveComponents.
Build LiveView: async data (assign_async), PubSub (check connected?), phx-change events, form components/modals/uploads, streams for lists, live_patch.
Principles for writing simple, maintainable Laravel/Livewire code. Use when writing Livewire components, tests, or Blade views. Focuses on avoiding over-engineering.
Produce a living architecture document as highly visual HTML that explains how the system actually works, why it was built this way, and how data flows and is shaped.
Navigate and load project living documentation for context from .specweave/docs/internal/. Use when implementing features and needing project context, referencing ADRs for design…
Runs LLM inference on CPU, Apple Silicon, and consumer GPUs without NVIDIA hardware. Use for edge deployment, M1/M2/M3 Macs, AMD/Intel GPUs, or when CUDA is unavailable — from…
Runs LLM inference on CPU, Apple Silicon, and consumer GPUs without NVIDIA hardware. Use for edge deployment, M1/M2/M3 Macs, AMD/Intel GPUs, or when CUDA is unavailable — from…
Automates LLDB debugging sessions with scripted breakpoint management and expression evaluation. Uses the LLDB Python SB API (lldb.SBDebugger, SBTarget, SBProcess) for…
Automatically applies when building LLM applications. Ensures proper async patterns for LLM calls, streaming responses, token management, retry logic, and error handling.
Production-ready patterns for building LLM applications. Covers RAG pipelines, agent architectures, prompt IDEs, and LLMOps monitoring.
Production-ready patterns for building LLM applications. Covers RAG pipelines, agent architectures, prompt IDEs, and LLMOps monitoring.
Production-ready patterns for building LLM applications. Covers RAG pipelines, agent architectures, prompt IDEs, and LLMOps monitoring.
Use when building LLM applications: prompt engineering, structured output, agents, RAG integration, memory management, or production deployment.
Expert LLM architect specializing in large language model architecture, deployment, and optimization.
Use when user needs LLM system architecture, model deployment, optimization strategies, and production serving infrastructure.
Use when user needs LLM system architecture, model deployment, optimization strategies, and production serving infrastructure.
LLM architecture, tokenization, transformers, and inference optimization. Use for understanding and working with language models.
Generate and execute terminal workflows using llm-box. Use when the user wants to automate multi-step terminal tasks, chain commands, fetch URLs, process data, create reusable…
Generate and execute deterministic terminal workflows using llm-box. Use when the user wants to automate multi-step terminal tasks, chain commands, fetch URLs and save results,…
Implement multi-layer LLM caching with exact match, semantic similarity, and provider-side prompt caching.
Multi-level caching strategies for LLM applications - semantic caching (Redis), prompt caching (Claude/OpenAI native), cache hierarchies, cost optimization, and Langfuse cost…
Implement or refactor a reusable LLM client layer for text-only or multimodal projects. Use when a project needs one final repo-local client module, OpenAI-compatible provider…
Cut LLM cost and latency with caching, model tiering, prompt diet, batching, and streaming UX. Use when the inference bill or response time needs engineering down.
Model the cost and latency of an LLM feature before it ships and surprises the bill. Use when asked to estimate LLM API costs, set a latency/token budget, decide which model tier…
Reduce LLM API costs without sacrificing quality. Covers prompt caching (Anthropic), local response caching, prompt compression, debouncing triggers, and cost analysis.
Use when you need to reduce LLM API spend, control token usage, route between models by cost/quality, implement prompt caching, or build cost observability for AI features.
Automate construction data processing using LLM (ChatGPT, Claude, LLaMA). Generate Python/Pandas scripts, extract data from documents, and create automated pipelines without deep…
Diagnoses LLM output failures including hallucinations, constraint violations, format errors, and reasoning issues.
当需要部署或训练 LLM/VLM 时使用；覆盖 vLLM OpenAI-compatible 服务、多模态输入限制、Qwen3.5 工具调用、thinking/reasoning 控制、CUDA Graph 策略，以及 ms-swift SFT/DPO/GRPO full training、数据校验、显存排错和训练检查。
LLM deployment and serving — vLLM, Ollama, TGI, llama.cpp. Model quantization, GPU optimization, API serving
Patterns for building LLM applications - prompt engineering, RAG pipelines, cost optimization, multi-model routing, and evaluation. Auto-triggers when working with AI/LLM code.
LLM evaluation harness for accuracy benchmarking. MMLU/HumanEval/MATH eval runners, model-graded scoring, prompt regression testing, and per-skill accuracy tracking.
Master comprehensive evaluation strategies for LLM applications, from automated metrics to human evaluation and A/B testing. — from whatamelon/AGENT-HARNESS-BASELINE
Implement comprehensive evaluation strategies for LLM applications using automated metrics, human feedback, and benchmarking.
Implement comprehensive evaluation strategies for LLM applications using automated metrics, human feedback, and benchmarking.
Implement comprehensive evaluation strategies for LLM applications using automated metrics, human feedback, and benchmarking.
Implement comprehensive evaluation strategies for LLM applications using automated metrics, human feedback, and benchmarking.
Implement comprehensive evaluation strategies for LLM applications using automated metrics, human feedback, and benchmarking.