Skills
98,284 packaged capabilities and playbooks for agents. Install any of them with one command.
Development
When calling LLM APIs from Python code. When connecting to llamafile or local LLM servers. When switching between OpenAI/Anthropic/local providers. When implementing retry/fallback logic for LLM calls. When code imports litellm or uses completion() patterns.
Use for any task or question involving Livewire. Activate if user mentions Livewire, wire: directives, or Livewire-specific concepts like wire:model, wire:click, invoke this skill. Covers building new components, debugging reactivity issues, real-time form validation, loading states, migrating from Livewire 2 to 3, converting component formats (SFC/MFC/class-based), and performance optimization. Do not use for non-Livewire reactive UI (React, Vue, Alpine-only, Inertia.js) or standard Laravel forms without Livewire.
Design and build UI with @loamui/core. The conductor for LoamUI work. It holds the philosophy and composition model, orders the work across the frontend-design, modern-css and modern-web-guidance skills, and runs the review before anything is reported. Use for any LoamUI build, design, theme or review task.
Compare two code versions for semantic equivalence via semi-formal tracing of both versions side-by-side. Trigger when the user shares a refactor, rewrite, migration, or A/B implementation and wants to confirm behavior is unchanged — "did I break anything", "is this equivalent", "are these equivalent", "semantically equivalent", "are these two implementations semantically equivalent", "check my refactor", "same behavior after the change?", "does my rewrite produce the same output", "switched from X to Y — same results?". SCOPE HARD RULE: requires two code versions (A and B). A single version for bug-finding uses logic-review; one version + a failing test uses logic-locate; explaining what one piece of code does uses logic-explain; codebase audit uses logic-health. Do NOT trigger for: single-version review, performance comparison, design-quality comparison, or "which is better-written" questions.
Establish MAVLink connections between MATLAB and PX4/ArduPilot autopilots. Use when connecting to a drone, flight controller, or autopilot via MAVLink protocol over UDP. Covers dialect setup, UDP transport, timer-based heartbeat, and client discovery. Use when: "connect to PX4", "MAVLink connection", "heartbeat", "ground control station", "GCS", "connect to ArduPilot", "drone communication", "mavlinkio", "SITL".
.NET MAUI view animations, custom animations, easing functions, rotation, scale, translation, and fade effects. USE FOR: "animate view", "fade in", "fade out", "slide animation", "scale animation", "rotate view", "translate view", "easing function", "custom animation", "animation chaining", "ViewExtensions animation". DO NOT USE FOR: gesture recognition (use maui-gestures), custom drawing (use maui-graphics-drawing), or static layout changes (use maui-data-binding).
Carry the current repository change through a verified GitHub pull request merge into the requested base, or the repository's remote default branch when none is named, then hand the merged branch to clean-branches so its worktree and local branch are cleaned up with the project's declared cleanup commands. Always use this skill for an actual PR merge or its post-merge cleanup, including requests to inspect, finish, land, or merge an existing PR; create a PR and merge it; preserve meaningful commits; or clean up an already-merged worktree. Report an existing merge instead of duplicating it. A sweep of a repository's accumulated branches and worktrees belongs to clean-branches.
Provides guidance for enterprise-grade RL training using miles, a production-ready fork of slime. Use when training large MoE models with FP8/INT4, needing train-inference alignment, or requiring speculative RL for maximum throughput.
LLM and ML model deployment for inference. Use when serving models in production, building AI APIs, or optimizing inference. Covers vLLM (LLM serving), TensorRT-LLM (GPU optimization), Ollama (local), BentoML (ML deployment), Triton (multi-model), LangChain (orchestration), LlamaIndex (RAG), and streaming patterns.
Integrate Modellix's unified API for AI image, video, and audio workflows. Use this skill whenever the user wants to generate or edit images, create or transform videos, synthesize speech, transcribe audio, clone a voice, do virtual try-on, or call any Modellix model API. Also trigger when the user mentions Modellix, model-as-a-service for media generation, or providers such as Qwen, Wan, Seedream, Seedance, Kling, Hailuo, MiniMax, Whisper, or CosyVoice through a unified API, or when they ask for a Modellix model's request schema, OpenAPI contract, or required input fields. Prefer modellix-cli (model get-schema, model run --wait, task download, doctor, model list) over hand-rolled REST polling whenever the CLI is available.
Build production-ready Node.js backend services with Express/Fastify, implementing middleware patterns, error handling, authentication, database integration, and API design best practices. Use when creating Node.js servers, REST APIs, GraphQL backends, or microservices architectures.
Handle FormData, file uploads, Blob, Uint8Array, and ReadableStream inputs in tRPC mutations. Use octetInputParser from @trpc/server/http for binary data. Route non-JSON requests with splitLink and isNonJsonSerializable() from @trpc/client. FormData and binary inputs only work with mutations (POST).
Essential patterns, idioms, and gotchas for writing Nushell code. Use when writing Nushell scripts, functions, or working with Nushell's type system, pipelines, and data structures. Complements plugin development knowledge with practical usage patterns.
Write, review, debug, and explain OCaml code and dune projects. Use whenever the user mentions OCaml, .ml/.mli files, dune, dune-project, opam, Alcotest, qcheck, shrinking, counterexamples, property-based testing, mutation testing, katas or exercises written in OCaml, or asks to convert JS/TS/PHP logic into OCaml. Trigger even if the user just pastes OCaml code with a compiler error and asks "what's wrong here", asks to set up a new dune project, or asks for the OCaml idiom for something they already know how to do in JS/TS. The user is an experienced JS/TS/PHP developer but a genuine OCaml beginner — explanations should assume strong general programming skill and zero OCaml-specific knowledge, never the reverse. DO NOT USE for F#, ReasonML/ReScript, Standard ML, or other ML-family languages unless the user is explicitly comparing them to OCaml.
Run local code quality checks covering a subset of OpenHarmony gate CI (copyright, CodeArts C/C++) plus additional local checks (pylint/flake8, shellcheck/bashate, gn format). Use before committing to reduce gate failures. Triggers on: /oh-precommit-codecheck, "门禁检查", "门禁预检", "检查代码", "run codecheck", "check code quality", "lint my code", "代码检查", or after completing code implementation. WHEN to use: before git commit, before creating PR, after modifying C/C++/Python/Shell/GN files, when gate CI fails with codecheck defects, or when you want to preview what gate will flag.
GPU-accelerates scientific Python on NVIDIA hardware and verifies that the result is correct and faster. Use for CUDA/GPU optimization; CPU-bound NumPy, SciPy, pandas, scikit-learn, NetworkX, scikit-image, vector-search, image-processing, graph, simulation, or file-I/O workloads; CuPy, cuDF, cuML, cuGraph, cuVS, cuCIM, KvikIO, Warp, Newton, Numba-CUDA, or RAFT questions; and profiling, memory-transfer, kernel, or multi-GPU bottlenecks. Also use when large data-parallel Python code is slow and GPU acceleration is a plausible option, even if the user does not name CUDA.
Optimize Decentraland scene performance. Scene limit formulas (triangles, entities, materials, textures, height per parcel count), object pooling, LOD patterns, texture optimization, system throttling, and asset preloading. Use when the user wants to optimize, improve performance, fix lag, reduce load time, check limits, or reduce entity/triangle count. Do NOT use for deployment (see deploy-scene).
Enables Flash Attention for transformer models using PyTorch native scaled_dot_product_attention (PyTorch 2.2+) or the flash-attn library, including multi-query attention, sliding window attention, and FP8 on H100 (FlashAttention-3). Covers profiling speedup, checking accuracy against a baseline, and troubleshooting install and GPU support errors. Use when training or running transformers on long sequences (over 512 tokens), when standard attention runs out of GPU memory, when attention is the inference bottleneck, when switching a PyTorch model to the flash backend, or when tuning attention on H100 GPUs. Not for CPU inference, V100 GPUs, or sequences under 256 tokens; consider xFormers for other attention variants.
Connects to Oxylabs remote headless browsers via Chrome DevTools Protocol (CDP) using Playwright or Puppeteer. Provides anti-detection, CAPTCHA handling, residential proxies, and geo-targeting built in. Use when browser automation needs remote execution, stealth capabilities, rendered pages, screenshots, PDFs, or complex JavaScript interaction.
Build with and use Pi, the minimal terminal coding harness. Use for installing Pi, configuring providers/models/settings/environment variables, creating Pi skills/extensions/packages/themes/prompt templates, embedding Pi through the SDK, integrating over RPC or JSON event streams, parsing sessions, running local models through the llama.cpp router, developing custom Pi providers and TUI components, or using ecosystem packages such as pi-subagents (delegation/orchestration), pi-mcp-adapter (MCP servers), pi-interview (interactive forms), and pi-web-access (web search, fetching, video understanding).
Run a tight interrogation of the need and the technical breakdown of a feature until you reach a shared plan, then hand the writing to planify-write-plan. Open questions grouped into independent batches, each with a recommended answer. Use when the user wants to plan a feature, break a need into easy-to-review tasks, stress-test a plan before coding, prepare a plan-mode session, or invokes /planify. Usage: /planify [demande]
Internal writer of the planify skill: formats the summary of a plan (produced by planify's interrogation) into two files under .task/<KEY>/<prefix>/ — a readable plan.md (tasks + acceptance constraints) and a dense tech.md. Invoked by planify through the Skill tool; does not trigger on its own and has no slash command. Asks no further questions and makes no decisions.
Help users integrate Portaly digital products checkout — list a creator's digital products and let buyers purchase one item or a custom bundle via Portaly's hosted checkout, with signed webhook callbacks. Also covers test mode — which test card to use, and why a test purchase sends no email and shows up in no revenue figure. Trigger when the user mentions Portaly digital products, selling courses/downloads/templates via their own site backed by Portaly, building a "powered by Portaly" storefront, bundle pricing of Portaly products, publishing or unpublishing a Portaly product (taking one off sale from their own site), or is troubleshooting a Portaly test payment, test card, sandbox order, or a missing order confirmation email.
EffectComposer setup and architecture for Three.js post-processing pipelines. Use when setting up multi-pass rendering, combining effects, creating custom passes, managing render targets, or building reusable effect stacks. Foundation skill for all post-processing work.
Generates PR reference XML with commit history and unified diffs between branches, with extension and path filtering. Use when creating pull request descriptions, preparing code reviews, analyzing branch changes, discovering work items from diffs, or generating structured diff summaries.
Load PROACTIVELY when starting work on an unfamiliar codebase or setting up a new project. Use when user says "help me understand this codebase", "onboard me", "what does this project do", "set up my environment", or "map the architecture". Covers codebase structure analysis, architecture mapping, dependency auditing, convention and pattern detection, developer environment setup, and documentation of findings for rapid productive contribution.
Expert skill for generating interactive HTML product prototypes. Use when the user needs to create product prototypes, UI interface prototypes, interactive HTML demo pages, product feature showcase pages, or mentions "原型", "prototype", "产品界面", "UI 原型", "交互原型", "演示页面" and the like.
Build real-time applications with PubNub pub/sub messaging. Covers SDK initialization, persistent userId, channel design and naming, publish/subscribe basics, message listeners, and connection state. Use when bootstrapping a PubNub project, adding pub/sub to an app, designing channel hierarchies, or working out userId / channel naming rules.
Solves single- and multi-objective optimization problems in Python with pymoo, using NSGA-II, NSGA-III, MOEA/D, SPEA2, RVEA, GA, DE and PSO. Covers custom problems (Problem, ElementwiseProblem, FunctionalProblem), constraint handling, mixed-variable problems, ZDT/DTLZ/WFG benchmarks, genetic operators, parallel evaluation, Pareto front visualization and multi-criteria decision making. Use when finding Pareto-optimal trade-offs between conflicting objectives. Use when defining a constrained or mixed-variable problem and choosing an evolutionary algorithm. Use when benchmarking algorithms on standard test problems. Use when customizing crossover or mutation operators. Use when picking one solution from a Pareto front. Not for gradient-based or convex solvers.
Anti-analysis countermeasure gateway. Orchestrates: packer identification → simple unpacking → advanced unpacking → deobfuscation. Sub-skills: [[re-packer-id]] [[re-unpack-simple]] [[re-unpack-advanced]] [[re-deobfuscate]] [[re-evasion]]. Triggers: unpacking, packer detection, packer identification, packer, UPX, VMProtect, Themida, anti-debugging, deobfuscation, junk code, unpack, anti-analysis.
Software reverse engineering core gateway (shared foundation). Orchestrates: initial survey → format parsing → decompilation → debugging/tracing → memory. Sub-skills: [[re-address-space]] [[re-triage]] [[re-format-pe]] [[re-format-elf]] [[re-format-macho]] [[re-imports]] [[re-ghidra]] [[re-ida]] [[re-radare2]] [[re-gdb]] [[re-x64dbg]] [[re-lldb]] [[re-tracing]] [[re-memdump]] [[re-windbg]] [[re-binaryninja]] [[re-emulation]] [[re-shellcode]] [[re-kernel]] [[re-game]] [[re-go]] [[re-rust]] [[re-plugin-dev]] [[re-hypervisor]] [[re-anti-cheat]], [[re-cpp-abi]], [[re-swift]], [[re-zig]], [[re-nim]], [[re-fp-runtime]], [[re-variant]], [[re-mips]], [[re-arm]], [[re-riscv]]. Triggers: static analysis, look at this program's logic, decompilation, reverse this binary, binary analysis.
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.