性能
性能分析与优化
serving-llms-vllm
majiayu000
Serves LLMs with high throughput using vLLM's PagedAttention and continuous batching. Use when deploying production LLM APIs, optimizing inference latency/throughput, or serving models with limited GPU memory. Supports OpenAI-compatible endpoints, quantization (GPTQ/AWQ/FP8), and tensor parallelism.
mamba-architecture
majiayu000
State-space model with O(n) complexity vs Transformers' O(n²). 5× faster inference, million-token sequences, no KV cache. Selective SSM with hardware-aware design. Mamba-1 (d_state=16) and Mamba-2 (d_state=128, multi-head). Models 130M-2.8B on HuggingFace.
gptq
majiayu000
Post-training 4-bit quantization for LLMs with minimal accuracy loss. Use for deploying large models (70B, 405B) on consumer GPUs, when you need 4× memory reduction with <2% perplexity degradation, or for faster inference (3-4× speedup) vs FP16. Integrates with transformers and PEFT for QLoRA fine-tuning.
rwkv-architecture
majiayu000
RNN+Transformer hybrid with O(n) inference. Linear time, infinite context, no KV cache. Train like GPT (parallel), infer like RNN (sequential). Linux Foundation AI project. Production at Windows, Office, NeMo. RWKV-7 (March 2025). Models up to 14B parameters.
application-performance-performance-optimization
rmyndharis
"Optimize end-to-end application performance with profiling, observability, and backend/frontend tuning. Use when coordinating performance optimization across the stack."
prep-with-circleback
superset-sh
Build a brief for an upcoming meeting from Circleback — the calendar event, the history with the same people and company, open commitments on both sides, and what is still unresolved. Use when the user asks to prep for a call, "what do I need to know before", who someone is, or where things stand with a customer or partner.
cost-optimization
wshobson
Optimize cloud costs through resource rightsizing, tagging strategies, reserved instances, and spending analysis. Use when reducing cloud expenses, analyzing infrastructure costs, or implementing cost governance policies.
ads-youtube
AgriciDaniel
YouTube Ads specific analysis covering campaign types, creative quality, audience targeting, and measurement. Evaluates video ad performance across skippable, non-skippable, bumper, Shorts, and Demand Gen formats. Use when user says "YouTube Ads", "video ads", "pre-roll", "bumper ads", "YouTube campaign", or "Shorts ads".
ai-shaped-readiness-advisor
deanpeters
Assess whether your product work is AI-first or AI-shaped. Score 5 competencies and recommend the next capability to build.
saas-economics-efficiency-metrics
deanpeters
Evaluate unit economics and capital efficiency for SaaS. Covers CAC, LTV, payback, margins, burn rate, Rule of 40, and magic number.
bazel-build-optimization
wshobson
Optimize Bazel builds for large-scale monorepos. Use when configuring Bazel, implementing remote execution, or optimizing build performance for enterprise codebases.
react-best-practices
laguagu
React and Next.js performance optimization guidelines from Vercel Engineering. This skill should be used when writing, reviewing, or refactoring React/Next.js code to ensure optimal performance patterns. Triggers on tasks involving React components, Next.js pages, data fetching, bundle optimization, or performance improvements.
action-queue
SimHacker
Sims-inspired task scheduling â queue actions, execute in order
agent-orchestration-multi-agent-optimize
benjaminasterA
"Optimize multi-agent systems with coordinated profiling, workload distribution, and cost-aware orchestration. Use when improving agent performance, throughput, or reliability."
application-performance-performance-optimization
benjaminasterA
"Optimize end-to-end application performance with profiling, observability, and backend/frontend tuning. Use when coordinating performance optimization across the stack."
agent-memory-systems
benjaminasterA
"Memory is the cornerstone of intelligent agents. Without it, every interaction starts from zero. This skill covers the architecture of agent memory: short-term (context window), long-term (vector s..."
arm-cortex-expert
rmyndharis
Senior embedded software engineer specializing in firmware and driver development for ARM Cortex-M microcontrollers (Teensy, STM32, nRF52, SAMD). Decades of experience writing reliable, optimized, and maintainable embedded code with deep expertise in memory barriers, DMA/cache coherency, interrupt-driven I/O, and peripheral drivers.
agent-orchestration-multi-agent-optimize
rmyndharis
"Optimize multi-agent systems with coordinated profiling, workload distribution, and cost-aware orchestration. Use when improving agent performance, throughput, or reliability."
bazel-build-optimization
rmyndharis
Optimize Bazel builds for large-scale monorepos. Use when configuring Bazel, implementing remote execution, or optimizing build performance for enterprise codebases.
code-refactoring-context-restore
rmyndharis
"Use when working with code refactoring context restore"
tech-contract-negotiation-patrick-munro
lawvable
"Guide to negotiating technology services agreements, professional services contracts, and commercial B2B transactions. Provides three-position frameworks (provider-favorable, balanced, client-favorable), deal-size tactics, objection handling templates, and concession roadmaps. Use when: (1) Developing negotiation strategies for SaaS, cloud, or managed services agreements, (2) Preparing position papers and fallback positions, (3) Responding to counterparty objections and demands, (4) Creating concession roadmaps that protect critical interests, (5) Assessing tactics based on deal value and leverage, or (6) Structuring balanced outcomes for liability, IP, payment, SLA, or warranty provisions."
nda-triage-anthropic
lawvable
Screen incoming NDAs and classify them as GREEN (standard), YELLOW (needs review), or RED (significant issues). Use when a new NDA comes in from sales or business development, when assessing NDA risk level, or when deciding whether an NDA needs full counsel review.
dbs-benchmark
Pluviobyte
dontbesilent 对标分析。用五重过滤法帮你找到值得模仿的对标，排除一切关于「我」的噪音。 触发方式：/dbs-benchmark、/对标、「帮我找对标」「我该模仿谁」 Benchmark analysis using dontbesilent's five-filter method. Trigger: /dbs-benchmark, "find me a benchmark", "who should I copy"
c-pro
rmyndharis
Write efficient C code with proper memory management, pointer