Skills
97,905 packaged capabilities and playbooks for agents. Install any of them with one command.
Development
Use this skill for Windows Forms Application needing Excel-like UI using the Syncfusion Spreadsheet Component. Trigger for creating, viewing, editing Excel (.xlsx, .xls) files; using formulas, filtering, sorting, or cell formatting. Also trigger when users reference spreadsheet files ("open xlsx", "load Excel file", "add Syncfusion spreadsheet"). Do NOT trigger for standalone file processing without UI components.
Use this skill for WPF Application needing Excel-like UI using the Syncfusion Spreadsheet Component. Trigger for creating, viewing, editing Excel (.xlsx, .xls) files; using formulas, filtering, sorting, or cell formatting. Also trigger when users reference spreadsheet files ("open xlsx", "load Excel file", "add Syncfusion spreadsheet"). Do NOT trigger for standalone file processing without UI components.
Build, review, debug, configure, or migrate Tailwind CSS projects using current Tailwind documentation. Use for Tailwind CSS v4.3 utilities, CSS-first configuration, theme variables, Vite/PostCSS/CLI/webpack/Turbopack setup, source detection, variants, responsive design, dark mode, Preflight, custom utilities, @apply/@reference, Prettier class sorting, and v3-to-v4 upgrades.
Use when consolidating, deduping, or reducing the number of CSS custom properties / Tailwind v4 design tokens in a globals.css file. Triggers on requests like "reduce tokens", "standardize design tokens", "too many tokens", "dedupe globals.css", "shrink @theme inline", "collapse similar colors", or any task whose goal is fewer tokens while keeping the same visual result. Covers @theme inline, --color-* aliases, :root/.dark/.dashboard-page blocks, dead-token analysis, value-similar collapse, single-use inlining, and the Tailwind v4 arbitrary-value escape syntax.
Optimizes LLM inference with NVIDIA TensorRT for maximum throughput and lowest latency. Use for production deployment on NVIDIA GPUs (A100/H100), when you need 10-100x faster inference than PyTorch, or for serving models with quantization (FP8/INT4), in-flight batching, and multi-GPU scaling.
Implement and debug Three.js interaction with Raycaster, pointer and touch coordinate conversion, hover/selection state, camera controls, dragging, keyboard input, and world-to-screen projection. Use when picking objects, adding OrbitControls or PointerLockControls, handling mouse/touch input, or fixing interactive 3D behavior. Triggers on: Three.js interaction, Raycaster, raycast, picking, click mesh, hover, object selection, OrbitControls, drag controls, pointer events, touch controls.
Create and debug Three.js shaders with ShaderMaterial, RawShaderMaterial, uniforms, varyings, GLSL vertex and fragment programs, texture sampling, procedural effects, onBeforeCompile extensions, shader chunks, instancing, and GPU performance checks. Use when writing custom GLSL, extending a built-in material, animating shader uniforms, or diagnosing shader compile, coordinate-space, and rendering issues. Triggers on: Three.js shader, ShaderMaterial, RawShaderMaterial, GLSL, uniforms, varyings, onBeforeCompile, fragment shader, vertex shader, fresnel, displacement, shader chunk.
Self-contained orchestration skill for writing high-performance TLE kernels and shipping TLE feature changes with reproducible validation. Use when the user wants to write/optimize TLE kernels, implement TLE API/verifier/lowering features, or debug TLE correctness/performance issues. Trigger on phrases like "write a TLE kernel", "optimize TLE operator", and "debug TLE local_ptr".
Provides guidance for PyTorch-native agentic RL using torchforge, Meta's library separating infra from algorithms. Use when you want clean RL abstractions, easy algorithm experimentation, or scalable training with Monarch and TorchTitan.
Trains large language models (2B-462B parameters) with NVIDIA Megatron-Core using tensor, pipeline, sequence, context, and expert parallelism, plus FP8 on H100 and MoE configuration for Mixtral-style models. Covers choosing TP/PP/DP/CP sizes, launching distributed training, tuning micro-batch size, and fixing low MFU, out-of-memory errors, and diverging loss. Use when training models above 10B parameters on NVIDIA A100/H100 GPUs, when setting up 3D parallelism for a LLaMA-style model, when configuring expert parallelism for MoE training, or when trying to raise MFU toward 40-47%. Use PyTorch FSDP, DeepSpeed, or HuggingFace Accelerate instead for models under 70B or simpler setups.
Provides guidance for mechanistic interpretability research using TransformerLens to inspect and manipulate transformer internals via HookPoints and activation caching. Use when reverse-engineering model algorithms, studying attention patterns, or performing activation patching experiments.
Use when debugging TypeScript/JavaScript bugs by tracing call chains, understanding unfamiliar codebases quickly, making architectural decisions, or reviewing code quality. Extract function signatures and JSDoc without full file reads, trace call hierarchies up/down, detect code smells, and follow data flow. Triggers on debugging, understanding codebase, architectural analysis, signature extraction, call tracing.
Master TypeScript's advanced type system including generics, conditional types, mapped types, template literals, and utility types for building type-safe applications. Use when implementing complex type logic, creating reusable type utilities, or ensuring compile-time type safety in TypeScript projects.
Guides Powermail 13+ form development for TYPO3, including form setup, finishers, validators, spam protection, ViewHelpers, PSR-14 events, TypoScript, email templates, conditional visibility, and extension points. Use when creating, debugging, or extending powermail forms, mail handling, validation, finishers, powermail_cond behavior, or tx_powermail data.
UE5.6-UE5.8 PCG building generation workflow for modular buildings, blockouts, facade rules, and runtime generation. Use when requests involve Procedural Content Generation (PCG), Shape Grammar, lot-based building spawn, deterministic random seeds, density/filter pipelines, or converting designer constraints into reusable PCG graphs.
UiPath API Workflow assistant — author, run, validate, package, publish, deploy, and troubleshoot JSON workflows for `uip api-workflow`. Load for ANY create/edit of a `Workflow.json` / API workflow project; with `evals/` present, tests come first (TDD). Covers Sequence, Assign, JavaScript, If (#Wrapper/#Then/#Else), ForEach, DoWhile, Break, TryCatch, Wait, Response, nested; files as JobAttachment refs via File to Base64 / Base64 to File (`$helpers.file.*`, `serializeData()`, `--input-file`/`--output-dir`); HTTP / IS connector activities via `uip api-workflow registry`. Operate: run, IS connections, pack/publish/deploy. Test/eval: `evals/<scope>/eval-sets/` datasets (exact-match, Evaluations panel); loop until green. Triggers on API workflows, project type \"Api\", JSON with `document.dsl`/`do[]`, those activity types, or file/base64 handling. Agent evals (`evals/eval-sets/`, no scope) & coded agents→uipath-agents. Flow & its evals→uipath-maestro-flow. .xaml/coded RPA→uipath-rpa. Coded Apps→uipath-coded-apps.
UiPath Human-in-the-Loop / HITL node authoring—building approval gates, escalations, write-back validation, and data enrichment checkpoints in Flow, Maestro, Case Management, or Coded Agents, plus a guided decision experience that captures reviewer corrections as decisions. NOT for managing, reassigning, or monitoring tasks at runtime (use uipath-tasks for that).
Generation of production-ready C# Unity code from functional specifications. Analyzes the existing codebase, chooses the right architectural pattern, generates code and tests. Triggers: /unity-code-gen, 'generer script Unity', 'creer composant', 'nouveau MonoBehaviour', 'creer ScriptableObject', 'generer code Unity', 'nouveau systeme Unity', 'ajouter feature Unity'. Use when the user asks to create a new script, component, system or feature in Unity. Produces C# files that follow the project's conventions, with associated NUnit tests.
Incremental and safe refactoring of Unity C# code. Detects code smells, proposes a plan prioritized by risk, and executes changes one by one with compilation verification. Triggers: /unity-refactor, 'refactor', 'code smell', 'clean code', 'dette technique', 'refactoring unity'. Produces restructured code without changing behavior.
Provides guidance on fine-tuning large language models with Unsloth, a library for faster, lower-memory training using LoRA and QLoRA, based on its official documentation (references/llms-txt.md). Use when setting up LoRA or QLoRA fine-tuning of an LLM with Unsloth, reducing GPU memory use during training, looking up Unsloth features or APIs, debugging Unsloth training code, or learning Unsloth best practices. Not for general model training with plain Hugging Face Transformers or PEFT without Unsloth.
Finds and fixes maintainability problems across a whole codebase or a user-specified scope, with explicit support for TypeScript type safety. Removes duplication, code smells, excessive complexity, poor composition, oversized files, confusing names, dead code, weak boundaries, type hacks, unsafe assertions, suppressed errors, and inconsistent patterns. Use when the user says "unsmell", requests codebase cleanup or refactoring, asks to remove duplication or smells, or wants a targeted area made simpler, safer, and more maintainable.
Velt Comments implementation patterns and best practices for React, Next.js, and web applications. Use when adding collaborative commenting features, comment modes (Freestyle, Popover, Stream, Text, Page), rich text editor comments (TipTap, SlateJS, Lexical), Apryse WebViewer (PDF/docx) comments, media player comments, chart comments, comments sidebar setup and customization (embed mode, floating mode, focused thread, V2 sidebar), sidebar filtering with accessModes for privacy, isAnnotationPrivate() visibility routing, CommentDialogActionService.isSubmitInFlight() for duplicate-submit guards, VeltCommentDialogAgentSuggestion primitives for AI suggestion accept/reject UIs, agent comment annotations via REST API (agent block with agentSource/agentId/executionId, agent-specific GET filters, suggestionAccepted/suggestionRejected client events, sourceType "agent" UI rendering), or binding Comment Bubble / Comment Dialog / Comment Tool wireframe slots via template variables (velt-data, velt-if, velt-class).
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.
Transform an AI agent into a disciplined software development partner with strong judgment, transparent decisions, proportionate verification, and craftsmanship. Use for software implementation, feature work, bug fixing, refactoring, optimization, migrations, and sustained collaboration inside a codebase. Do not use merely because a task involves an AI agent; use a dedicated agent-design skill when the primary problem is agent architecture.
Configure and optimize Vite 7 for React projects. Covers build tooling, dev server, plugins, HMR, chunk splitting, Environment API, and Rolldown integration. Use when setting up Vite, configuring builds, optimizing bundles, managing plugins, or troubleshooting dev server. Triggers on vite, vite config, vite plugin, HMR, dev server, build optimization, chunk splitting, rolldown, vite proxy, environment api, rolldown-vite.
Vue bindings for TanStack Router: RouterProvider, useRouter, useRouterState, useMatch, useMatches, useLocation, useSearch, useParams, useNavigate, useLoaderData, useLoaderDeps, useRouteContext, useBlocker, useCanGoBack, Link, Navigate, Outlet, CatchBoundary, ErrorComponent, Html, Body. Vue-specific patterns with Ref<T> returns, defineComponent, h() render functions, provide/inject, and computed refs.
Logs and tracks machine learning experiments with Weights & Biases (W&B, wandb): metrics, hyperparameters, checkpoints, sweeps, artifacts with lineage, model registry, custom charts and shareable reports. Includes integrations for PyTorch, HuggingFace Transformers, PyTorch Lightning and Keras/TensorFlow. Use when logging training runs and comparing them across configurations, running automated hyperparameter sweeps, versioning datasets and models as artifacts, registering model versions, or sharing results with a team workspace. Use when setting up wandb.init for offline or unstable-connection training. Not for general data versioning outside ML runs.
Plan and coordinate approved multi-task work from prose, issues, repository evidence, or an explicitly selected Project by grouping selected issues into coherent PRs, delegating implementation in reusable task worktrees, and delivering verified draft PRs with required native stacks for dependent PRs. Never merges.
Parallel blind search from multiple independent angles (by structure/files, by content/semantics, by git history/time, by entity/domain, by error paths, alternative-storage-paths, config/realm dimensions, etc.), then a BARRIER at the merge step that dedups against a persistent SEEN set and lists what no angle covered. Each searcher is kept ignorant of what the others found in the same round. Use when one search strategy is known to be incomplete: "find all X", security review, "where can this break?", data flow tracing. Invoke via /work-sweep.
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.