Skills
96,770 packaged capabilities and playbooks for agents. Install any of them with one command.
Development
Guides technical support engineering—customer ticket investigation, reproduction, log and API analysis, root-cause isolation, workaround communication, engineering escalation with evidence, and knowledge-base fixes for product bugs and integration issues. Use when debugging a customer-reported issue, writing a repro for engineering, analyzing API errors, drafting technical replies, or improving support runbooks—not for CS program design, renewals, or billing ops (customer-ops-specialist), production incident command (incident-management-engineer), building product features (fullstack-software-engineer), or company-wide crisis statements and launch announcements (communication-lead), or exec/VIP and community escalation program design (community-executive-escalations-program-manager). Product how-to, macros, and ticket triage without deep debugging: product-support-specialist.
Design and audit support-to-engineering escalation workflows covering severity, reproduction, logs, customer impact, ownership, SLAs, bug intake, incident boundary, feature requests, known issues, engineering triage, workaround, customer updates, closure proof, and feedback loops. Use when support needs reliable engineering help without noisy or invisible escalations.
Integrate new DEX aggregators, swappers, or bridge protocols (like Bebop, Portals, Jupiter, 0x, 1inch, etc.) into ShapeShift Web. Activates when user wants to add, integrate, or implement support for a new swapper. Guides through research, implementation, and testing following established patterns. (project)
Tailwind CSS v4 guidance: utility-first patterns, @theme directive for color/spacing/font config, @source for content paths, dark mode, responsive design, oklch colors, custom variants, v4 vs v3 differences. Use when user writes Tailwind utility classes, configures @theme or @source in CSS, asks about v4 syntax, or styles components.
Build full-stack React apps with TanStack Start — server functions, type-safe routing, loaders, middleware, SSR/streaming, and deployment patterns. Use when working on TanStack Start apps, server functions, TanStack Router, or any gremlin-cms development.
Document and prioritize a technical debt backlog with business impact, effort estimates, and resolution strategy. Use when asked to audit technical debt, create a debt register, prioritize tech debt for a quarter, document architectural shortcuts, or build a debt reduction roadmap. Produces a structured technical debt register covering debt inventory by category, business impact per item, effort and priority scores, top-item resolution plans, and a quarterly debt reduction roadmap.
Code review and design support based on the "Tell, Don't Ask" principle. It converts patterns that query an object's state and decide externally into patterns that tell the object directly. It promotes designs that strengthen encapsulation and concentrate responsibility in the object that holds the data. Use when code review, new implementation, or refactoring requires fixing getter abuse or Feature Envy. Target languages: Java, Kotlin, Scala, TypeScript, Python, Ruby, Go, Rust. Triggers: OOP design requests such as "I want to reduce getters", "improve encapsulation", "fix Feature Envy", "I want objects to take on responsibility", "Law of Demeter".
Invoked by /tidy-project. Read-only, four-phase audit of PROJECT STRUCTURE — directory and file layout plus statically auditable config, dependency, and build hygiene (nested codebases, missing lockfiles, committed build artifacts, over-nesting, misplaced files, config sprawl, vestigial scaffolding, hardcoded environment values, convention drift). Code-semantics concerns are out of scope. Produces a single consolidated execution plan at `docs/YYYY-MM-DD-tidy-project-plan.md` (or repo root if no `docs/`); modifies no source files. Skips traversing dot-directories and build/vendor output, though it may still flag their tracked presence via `git ls-files`. Does not auto-activate on natural-language requests.
Guidance for implementing tensor parallelism in PyTorch, including ColumnParallelLinear and RowParallelLinear layers. This skill should be used when implementing distributed tensor parallel operations, sharding linear layers across multiple GPUs, or simulating collective operations like all-gather and all-reduce for parallel computation.
Diagnose unstable or underperforming training runs from logs, curves, configs, and samples. Use when a run diverges, plateaus, collapses, overfits, underfits, behaves strangely, or needs debugging across optimization, regularization, and data pathologies.
TypeScript and JavaScript expert with deep knowledge of type-level programming, performance optimization, monorepo management, migration strategies, and modern tooling. Use PROACTIVELY for any TypeScript/JavaScript issues including complex type gymnastics, build performance, debugging, and architectural decisions. If a specialized expert is a better fit, I will recommend switching and stop.
UI/UX design intelligence. 50 styles, 21 palettes, 50 font pairings, 20 charts, 9 stacks (React, Next.js, Vue, Svelte, SwiftUI, React Native, Flutter, Tailwind, shadcn/ui). Actions: plan, build, create, design, implement, review, fix, improve, optimize, enhance, refactor, check UI/UX code. Projects: website, landing page, dashboard, admin panel, e-commerce, SaaS, portfolio, blog, mobile app, .html, .tsx, .vue, .svelte. Elements: button, modal, navbar, sidebar, card, table, form, chart. Styles: glassmorphism, claymorphism, minimalism, brutalism, neumorphism, bento grid, dark mode, responsive, skeuomorphism, flat design. Topics: color palette, accessibility, animation, layout, typography, font pairing, spacing, hover, shadow, gradient. Integrations: shadcn/ui MCP for component search and examples
Use when running UnoPim tests, linting, static analysis, or asset builds, or when asked whether a change is ready to be called done. Trigger phrases include “run tests,” “pest,” “pint,” “larastan,” “phpstan,” “playwright,” “build assets,” and “is this done.”
Convert ambiguous user requests into structured USDM requirements documents. Decomposes requirements into Requirement -> Reason -> Description -> Specification hierarchy. Integrates with GitHub Issues, Asana, and Jira tickets as input sources. Use when: create requirements, write requirements document, USDM, decompose requirements, requirements definition, 要件定義, 要件を整理, 要件分解.
This skill should be used when the user's request or requirement is ambiguous and needs iterative questioning to become actionable. Trigger on "clarify requirements", "refine requirements", "요구사항 명확히", "요구사항 정리", "뭘 원하는 건지", "make this clearer", "spec this out", "scope this", "/clarify". Turns vague inputs into concrete specs. For strategy blind spots use unknown; for content-vs-form reframing use metamedium.
Vercel KV, the Redis-compatible key-value store powered by Upstash, in Next.js and other Vercel applications. Covers @vercel/kv setup and environment variables (KV_REST_API_URL, KV_REST_API_TOKEN), key and JSON value operations, TTL and expiration, caching patterns for pages, APIs, and data, session and auth token storage, rate limiting, TypeScript usage from the edge runtime, serverless functions, and Server Actions, plus migrating from Cloudflare KV. Use when adding Vercel KV to an app, building a cache or rate limiter, storing sessions or temporary data with a TTL, choosing strongly consistent Redis-compatible storage over eventually consistent alternatives, or debugging KV_REST_API_URL not set, rate limit exceeded, or JSON serialization errors.
Guides web application development—browser-based products spanning UI, HTTP APIs, sessions and cookies, routing (SPA and SSR), forms, file uploads, and web-specific security (CSRF, CORS, CSP, XSS prevention) on stacks such as React/Next.js, Vue, or similar with Node/Python/Ruby backends. Use when building or maintaining a web app, implementing login flows, server-rendered or hybrid pages, REST/GraphQL consumption, or debugging browser–server issues—not for mobile native apps, CLI tools, or data pipelines. For generic full-stack feature delivery use fullstack-software-engineer; for front-end-only work use senior-frontend-software-engineer; for deploy/CI use devops. For design-to-code UI implementation (tokens, states, Storybook) without auth/CORS focus, use ui-software-engineer.
Researches current web and external technical information through source-backed discovery, retrieval, and repository, document, or media inspection. Use this skill when the user needs current facts, docs, articles, URLs, repositories, PDFs, videos, comparisons, or version-specific library/API/framework evidence. Do not use this skill alone for interactive browser tasks such as login, forms, or UI-state inspection; use an available browser automation capability.
Write or overhaul a project's README.md, drawing on patterns from the most-starred GitHub readme-template repos and guides (othneildrew/Best-README-Template, the standard-readme spec, makeareadme.com, awesome-readme). Use this whenever the user asks to write, create, improve, restructure, or "make my README look better/more professional", asks for a README for a library/CLI/API/app, asks about badges/shields, a table of contents, "getting started" or "installation" sections, a Support or Project Status section, or wants their GitHub profile README (the special username/username repo) polished. Covers picking the right template variant (library, CLI tool, application, monorepo, or profile README), section ordering, badge selection, and common mistakes that make READMEs look templated or get skipped by readers.
Use before wt-work to verify prepared context is executable as wt development work, slice it, author TaskDocument bodies via wt-writing-tasks, choose execution shape and workflow policy, and hand off an exact launch target. For vague work that still needs thinking, route to leaf-work and stop.
Add a post-processing effect stack to a React Three Fiber scene — bloom, depth of field, vignette, chromatic aberration, noise/grain, tone mapping, SSAO, and custom effects via @react-three/postprocessing. Use when the user wants the render to look more cinematic, polished, or "premium," or asks for glow/bloom/DOF/film-grain style effects
Analyze repository structure and generate or update standardized AGENTS.md files that serve as contributor guides for AI agents. Supports both single-repo and monorepo structures. Measures LOC to determine character limits and produces structured documents covering overview, folder structure, patterns, conventions, and working agreements. Update mode refreshes only the standard sections while preserving user-defined custom sections. Use when setting up a new repository, onboarding AI agents to an existing codebase, updating an existing AGENTS.md, or when the user mentions AGENTS.md.
Design multi-objective e-commerce product ranking combining relevance, conversion, and business metrics. Use this skill when the user needs to build a product ranking system beyond text relevance, balance relevance with commercial objectives, or implement learning-to-rank — even if they say 'product sorting', 'search result ranking', or 'how to rank products'.
Implement LDA topic modeling to discover latent topics in document collections. Use this skill when the user needs to extract topics from a text corpus, categorize documents by theme, or explore thematic structure — even if they say 'what are the main topics', 'topic extraction', or 'document clustering by theme'.
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.