性能
性能分析与优化
dotnet-code-review
Olino3
Forge is a marketplace for a Claude Code Plugins
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.
stage-plan
cxyliuy
Convert vague or high-level requests into a concrete engineering plan with goals, constraints, steps, success criteria, and validation. Stop and wait for human confirmation.
code-reviewer
pierreribeiro
Code Reviewer persona for technical code analysis. ACTIVATE when messages contain review code, analyze code, code quality, best practices, refactor, improve code, check code, code review, look at code, feedback on code, or constructive feedback requests.
swiff-ios
Narenreddy2302
Expert assistant for the Swiff iOS subscription and expense management application. Use when working on Swiff iOS features, architecture, SwiftUI/SwiftData development, widgets, notifications, subscriptions, expenses, group billing, security, accessibility, or performance optimization. Provides comprehensive knowledge of MVVM architecture, service layer patterns, and iOS best practices.
cloud-architect
Olino3
"Multi-cloud architecture and cloud-native design patterns. Analyzes infrastructure requirements, designs scalable cloud solutions across AWS, Azure, GCP, and recommends cloud-native patterns. Evaluates architectural trade-offs, cost optimization, and migration strategies."
proactive-agent
cgk-platform
'Transform AI agents from task-followers into proactive partners that anticipate needs and continuously improve. Now with WAL Protocol, Working Buffer, Autonomous Crons, and battle-tested patterns. Part of the Hal Stack 🦞'
ad-pre-launch-scoring
tonyflo79
Probabilistic evaluation and variant prioritization for ad testing sequences. Use after the variant matrix (A09) is assembled and you need to determine which variants to test first. Scores variants with probability ranges (not false precision) and assigns Tier 1/2/3 priority for testing sequence. Every score traces to evidence from A01 intelligence, A06 Arena results, benchmarks, or historical data. The goal is prioritization, not perfection — tier assignment matters, decimal precision does not. Trigger when users mention ad scoring, variant prioritization, which ads to test first, pre-launch evaluation, or predicting ad performance. Requires A09 variant matrix.
techno-humanist-prophet
shadowcz007
从「传入的简报 + 用户记忆」构成的上下文中挑选素材与洞察，用乔布斯与凯文·凯利的双重视角深度解读，产出科技人文主义推文。
LeetCode
imoonkey
```
deep-research
janjaszczak
Conduct deep, source-grounded research with freshness checks and citations, then synthesize an actionable answer. Use for time-sensitive, niche, or high-stakes topics; when the user asks to “zweryfikuj”, “zrób research”, or when uncertainty is high.
animate
Olino3
"Zero-config animations for React, Vue, Solid, Svelte, Preact with @formkit/auto-animate (3.28kb). Prevents 15 errors."
supabase-postgres-best-practices
AidanGio
Postgres performance optimization and best practices from Supabase. Use this skill when writing, reviewing, or optimizing Postgres queries, schema designs, or database configurations.
dean-large-scale-systems
copyleftdev
Design large-scale systems using Jeff Dean's pragmatic approach. Emphasizes performance intuition, building reliable systems from unreliable components, and solving problems at Google scale. Use when building infrastructure that must handle millions of QPS, petabytes of data, or global distribution.
jump-trading-fpga-hft
copyleftdev
Build trading systems in the style of Jump Trading, the high-frequency trading firm pioneering FPGA-based trading. Emphasizes hardware acceleration, network optimization, and nanosecond-level execution. Use when building FPGA trading systems, network-optimized infrastructure, or ultra-low-latency order execution.
baby-skill
henryjohnbrand
Core baby intelligence skill. Records your baby's reactions, preferences, routines, and what works to soothe, entertain, and engage them. Builds a living profile that grows with your baby. Works across all ages 0-3 years. Self-learning — every interaction makes it smarter about YOUR specific baby. Not medical advice. Your parenting co-pilot.
opinion-critic
dtsvetkov1
Provides critical feedback and alternative architectural perspectives on code changes. Use when you want to explore different ways to implement a feature or find potential flaws in the current plan.
browser-auth-cdp
Aden232003
Connect agent-browser to an existing logged-in Chrome session via Chrome DevTools Protocol (CDP). Use when you need to automate browsers with existing login sessions (X/Twitter, Skool, Gmail, etc.) without re-authenticating. Triggers on "logged in browser", "existing session", "Chrome CDP", "bypass login", or "use my Chrome".
speed-loop
toomingos
"Systematically optimizes a data processing flow through iterative speed loops. Use when the user wants to improve performance of a data pipeline, query, or processing endpoint. Invoke with the flow to optimize: /speed-loop path/to/file or /speed-loop describe the flow"
skill:firebase-auth - Firebase Authentication
Olino3
Forge is a marketplace for a Claude Code Plugins
pillar-cornerstone-writer
Hasim-hannover
Write conversion-focused cornerstone articles for WordPress pillar categories (Strategie, SEO, Tracking, CRO, Performance). Use when the user asks for long-form category lead articles, decision-maker content, storytelling with business outcomes, SEO metadata, URL slug recommendations, or internal-link plans.
swiftui-patterns
affaan-m
SwiftUI architecture patterns, state management with @Observable, view composition, navigation, performance optimization, and modern iOS/macOS UI best practices. Use when building or reviewing SwiftUI views, @Observable state, navigation, or render performance.
go
kprsnt2
Best practices for Go development including idiomatic patterns, concurrency, and error handling.
mode-awareness
jamelna-apps
When explicitly switching modes or when mode detection shows low confidence. Provides guidance on optimal approach for each detected mode (debugging, performance, feature, refactor, exploration, infrastructure).