Skills
101,323 packaged capabilities and playbooks for agents. Install any of them with one command.
All
Use when drafting a grant or challenge proposal for a radiology or medical AI project, including a Korean government industry-academia plan. Structures significance, innovation, approach, aims, milestones and consortium roles, keeping claims evidence-based and executable.
Pressure the facts behind a claim before a decision closes — counter-question with a concrete example, capped attempts, escape hatches, evidence hierarchy — and route what comes out to the loaded node's decisions or backlog. Manually triggered, and also the method any other interviewing tool cites by path.
Automates Google Tag Manager API setup including googleapis installation, OAuth credential creation, token management, and prerequisites validation. Use when users need to "set up GTM API", "configure GTM API access", "get GTM OAuth credentials", "install googleapis", or encounter authentication errors. Handles complete technical setup from dependency installation through API connection verification.
Generate thorough, interactive HTML testing checklists — a real test plan organized into end-to-end flows, with pass/fail/blocked step states, a progress bar, filter and search, and copyable, syntax-highlighted command snippets. Always ends with a Submit button that returns every step's state and notes for failure triage and fixes. Use whenever the user wants to test, QA, verify, or validate a change, feature, release, PR, or bug-fix batch. Embedded snippets pass a redaction step so credential-shaped values never land in the artifact.
Identify and remove AI writing traces in Chinese text, making copy, articles, reports, emails, social media content, and technical explanations natural, specific, and genuinely human-written without fabricating facts or changing the original meaning. Use when the user asks to "remove the AI flavor", "reduce the machine feel", "make it more natural/more colloquial/more like me", or needs to review Chinese expression issues such as jargon, abstract verbs, vague elevation, marketing tone, piled-up parallelism, or mechanical paragraphing.
Implements GDPR (EU 2016/679) technical and organizational measures — privacy by design/default, DPIAs, data subject rights management, 72-hour breach notification, and cross-border transfer mechanisms (SCCs, BCRs, adequacy). Use when designing or auditing GDPR controls, building a DPIA, handling data subject access/erasure requests, or assessing cross-border data transfers.
Integrate Arcjet Guard into Python OpenAI Agents — wrap FunctionTool / function_tool with guard_tool and read a caller-owned session or conversation id via openai_agents_context. Use when asked to add Arcjet to openai-agents, Runner.run, rate limit those tools, screen inbound messages, or block prompt injection / PII. This is Python openai-agents, not the JS @openai/agents adapter.
JavaScript/TypeScript SDK for inference.sh - run AI apps, build agents, integrate with all models. Package: @inferencesh/sdk (npm install). Full TypeScript support, streaming, file uploads. Build agents with template or ad-hoc patterns, tool builder API, skills, human approval. Use for: JavaScript integration, TypeScript, Node.js, React, Next.js, frontend apps. Triggers: javascript sdk, typescript sdk, npm install, node.js api, js client, react ai, next.js ai, frontend sdk, @inferencesh/sdk, typescript agent, browser sdk, js integration
Reconstruct a published mechanistic, QSP, PK/PD, or ODE model in Jinkō from papers, supplements, equations, diagrams, tables, and optional source code. Use whenever the requested result is an executable paper reproduction, including when no code exists or a scanned PDF needs OCR. Produces a validated Jinkō model when evidence is sufficient, or a documented blocked or failed outcome, and always produces a reproduction report with explicit fidelity, confidence, assumptions, discrepancies, and numerical comparison where the paper supplies a target. Use jinko-task-from-nonmem when NONMEM is the primary implementation source.
Turn a short message - an announcement, a stat, a quote, a manifesto line - into a 20-40 second kinetic-typography video where animated words are the only actor on screen. Use when the user says "kinetic text video", "animated text video", "typography video", "make this quote/stat into a video", "text animation for this announcement", or gives you a few lines of copy and wants them turned into motion.
Inventory and topology of the monitored Kubernetes estate via KubeSense MCP — clusters, nodes, pods, workloads, detected issues, infra failures (OOM/CrashLoop/probe/scheduling), and recent deploys/scaling changes. Use for "what is running", "what is broken at the k8s layer", and "what changed".
Query spans and distributed traces via KubeSense MCP — latency percentiles, error rates, the exact trace field catalog (service, role, method, status_code, resource — not app_service/kind/subtype/return_code), duration units, and the distributed-trace waterfall for root-causing which hop broke or burned the time.
Interview the user for the raw material their posts are made of. Builds a lasting Story Bank of roles, numbers, turning points, scars and positions, or runs a focused interview that turns one topic into a post spine. Use when a draft has nothing concrete to draw on, or the user says interview me. Not for learning how they write (use linkedin-humanizer --mode profile).
Lip-sync a face to a specific audio track on RunComfy via the `runcomfy` CLI. Routes across ByteDance OmniHuman (audio-driven full-body avatar from a portrait + audio), Sync Labs sync v2 / Pro (state-of-the-art mouth sync onto a video), Kling lipsync (audio-to-video and text-to-video with synced speech), and Creatify lipsync. The skill picks the right endpoint for the user's actual intent — portrait still + audio (avatar-style), source video + audio (mouth-swap on existing footage), or generate-and-sync from a script. Triggers on "lip sync", "lipsync", "make this video speak", "match audio to mouth", "dub video", "sync lips to voice", "Sync Labs", "voiceover sync", or any explicit ask to drive a face's mouth from an audio track.
Create a narrative guide to a codebase or feature in the style of Knuth's Literate Programming — code and prose interwoven as a single essay, ordered for human understanding rather than compiler needs. Use when the user asks to 'explain this codebase as a story', 'write a literate guide', 'create a narrative walkthrough', 'tell the story of this code', 'Knuth-style documentation', 'weave a guide for this feature', or when they want deep, readable documentation that treats the program as literature. Also trigger when someone wants a document that a thoughtful reader could follow from start to finish and come away understanding both WHAT the code does and WHY every design choice was made.
Expertise in LLVM-based code obfuscation techniques including OLLVM, control flow flattening, string encryption, virtualization, and anti-analysis methods. Use this skill when working on code protection, anti-reverse engineering, or implementing custom obfuscation passes.
Build maps in React with mapcn, themeable shadcn-style map components built on MapLibre GL and Tailwind CSS. Use when the user wants to add, customize, or theme a map: markers, popups, tooltips, routes, arcs, GeoJSON or choropleth layers, clusters, a globe, custom basemap styles or light/dark theming, or a full map page (analytics dashboard, delivery tracker, store locator) in a React, Next.js, or Vite project using shadcn/ui or Tailwind. Also use when they ask where a map would fit in their app.
Find the right market, extract real customer pain, and validate demand before building an offer. Use when you have an idea but no market, the niche feels too broad, or the offer is not converting. Generates micro-niches, scores them, extracts pain language, and creates a validation test plan — outputs MARKET_RESEARCH.md.
Terraform modules for deploying self-managed Materialize on AWS (EKS), Azure (AKS), and GCP (GKE). Covers networking, Kubernetes clusters, managed PostgreSQL, object storage, the Materialize operator, and Materialize instance configuration. Use this skill whenever the user asks about deploying self-managed Materialize with Terraform, setting up Materialize infrastructure on any cloud, configuring EKS/AKS/GKE for Materialize, troubleshooting a self-managed deployment, upgrading Materialize instances, customizing Terraform variables for Materialize, or working with the materialize-terraform-self-managed repository. Also trigger when the user mentions Karpenter node pools for Materialize, persist or metadata backend URLs, the Materialize operator Helm chart, CRD versions (v1alpha1 vs v1), rollout strategies, or NVMe swap configuration for Materialize nodes.
Generate images and videos through the currently installed Media.io CLI. Use for text-to-image, image-to-image, text-to-video, image-to-video, reference-to-video and published workflows. Select the model from the bundled static catalog (`references/model-catalog.md`), which is generated from the production registry; only fall back to `mediaio model list` on the triggers listed in the discovery guardrail. Copy every `job_type` byte for byte — some contain a literal space, and display names often do not match the identifier. Effects may be used only when their parameters have been independently verified because the current CLI exposes `effect list` but not `effect get`. Always confirm the parameter schema with `model get` before submission. Submitting spends the user's credits, but the CLI stays quiet about the amount unless asked; surface the cost with `--show-credit`, and get an explicit approval first, only when the user is cost-sensitive or has raised credits, price or balance. When the balance runs short, or the user wants their task history or their uploaded files, hand over the product-page link printed by `mediaio link get` — never write or edit a media.io URL yourself. Use the human-readable discovery output, upload local files before generation, submit with `generate create`, wait with the separate `generate wait` command, and retrieve result files with `generate download` instead of reproducing signed result URLs. On hosts that sandbox local command networking, the first networked `mediaio` or `curl` Shell/Bash tool call must request host approval before process launch. Never probe network availability by running it in the default sandbox.
Solana memecoin trading analysis and execution support. Use when analyzing tokens, detecting rugs, finding alpha, or planning trades on pump.fun, Raydium, Jupiter. Covers: token metrics, liquidity analysis, holder distribution, entry/exit signals, position sizing, degen strategies.
Use when someone wants to move an app from paywall placements to Flow Builder flows — "migrate my placements to flows", "switch my paywalls to flows", "we have 40 placements on paywalls and want flows", or a bulk request across many apps or paywalls. Creates NEW flow placements alongside the existing paywall ones and hands back the old→new mapping plus the SDK call change. Never converts a placement in place — the backend refuses a type change. Complements `flow-generator` (which builds and publishes flows) and `adapty-integration` (which owns the app-side call sites).
Production-grade mobile application UI design based on Refactoring UI principles. ALWAYS activate for: Flutter app, mobile app, iOS app, Android app, mobile UI, app screens, mobile navigation, bottom sheets, mobile forms, touch targets, mobile typography, app color scheme, mobile cards, list views, mobile modals, tab bars, app bars, floating action buttons. Provides mobile design workflow, touch-optimized spacing, mobile typography scale, platform-aware patterns. Turkish: mobil uygulama tasarımı, Flutter tasarım, uygulama arayüzü, mobil UI, telefon uygulaması, Android tasarım, iOS tasarım. English: app design, mobile interface, touch-friendly, native feel, mobile UX.
Make a web screen feel native inside a phone app — a paywall, an onboarding or quiz, an offer, an update screen, any screen rendered in a WKWebView / Android WebView or shipped as a Superwall surface. The rules a native screen follows and a web page doesn't: press feedback on every control, hover gated to real pointers, 16px inputs and the keyboard, motion with real easing and durations, size-aware type, worst-case content, dark mode from the host, iOS vs Android conventions, and the audit before it ships. Use when building or reviewing any screen of that kind for iOS or Android, with or without a design reference, or when a screen "feels like a website" on a phone. For the Superwall framework's own mechanics (insets, routing, products, push) read superwall-framework; for React Native / Expo screens use the platform's own components instead.
Anti-slop design skill for BUILDING React Native / Expo apps (iOS, Android, web). Commits to an App Read and Nav Read before writing code; ships screens, navigation, and motion that do not look AI-generated. Use when creating new screens, features, or apps in React Native. For auditing an existing app use mobile-design-review; for upgrading existing screens use mobile-redesign-screen or mobile-redesign-app; for navigation planning alone use mobile-nav-plan; for tokens/theming use mobile-design-system.
Add new/custom AI models to opencode.json. Use proactively for bleeding-edge releases, non-standard providers, self-hosted models, or custom endpoints. Examples: - user: "Add the new Claude 4.5" → websearch for API specs, add provider entry with baseUrl, verify model ID format - user: "Use my local Ollama instance" → configure custom provider with http://localhost:11434, set model ID format - user: "Configure this OpenAI-compatible proxy" → add provider with custom baseUrl, set apikey env var, verify compatibility - user: "Model X just released, add it" → research provider documentation, find model ID and capabilities, add to config
Generate images with Google Nano Banana 2 (Gemini-family flash-tier text-to-image) on RunComfy — bundled with the model's documented prompting patterns so the skill gets sharper output than naive prompting against the same model. Documents Nano Banana 2's strengths (rapid iteration, in-image typography rendering, predictable framing, optional web-grounded context), the resolution-tier pricing, the safety-tolerance dial, and when to route to Nano Banana Pro / GPT Image 2 / Flux 2 / Seedream instead. Calls `runcomfy run google/nano-banana-2/text-to-image` through the local RunComfy CLI. Triggers on "nano banana", "nano-banana-2", "nano banana 2", "google image gen", "gemini image", or any explicit ask to generate with this model
Verify Next.js runtime behavior after editing app code. Use this skill to confirm a change actually works in a running app—not just that it compiles or type-checks. Combines /_next/mcp (Next.js's view) with agent-browser (the browser's view). Requires a running `next dev`.
Convert a local skill into a Claude Code optimized shape — an AUDIT-then-rewrite that ties every change to real Claude skill-runtime behavior, treating FRONTMATTER as the highest-leverage routing surface. Inventories the skill, checks it against the runtime reference (Claude routes and budgets skills from metadata first), writes a per-skill DAG rewrite plan with source-backed guardrails, and — only when asked — rewrites frontmatter, body, and references for cheaper routing and safer execution. Defaults to PLAN mode, never a blind rewrite, uses skill-native frontmatter only, and preserves structural content (routing tables, guardrail rules) rather than stripping it as bulk. Use to migrate or restructure a SKILL.md for Claude's skill runtime.
Write the brains of an app's onboarding or web2app funnel — every question, its answers, the order, the branching, the statements and explainer screens between them, image descriptions and the meaning of each transition — so the flow primes the user to finish and to buy, and collects the answers that predict revenue (from ARPU data across 411 apps). Starts from an App Store URL (and the website, if there is one), researches the app, and hands over a spec that ships fastest as a Superwall surface (the superwall framework: one page per screen, answers sent as user attributes). Design, UI and code agnostic; no paywall, pricing or checkout. Use when the user asks \"what should we ask in onboarding\", \"write our onboarding questions\", \"design a quiz/survey funnel\", \"which questions predict who pays\", \"make our onboarding convert\", or wants inspiration for onboarding screens.
Use the public `orca` CLI to operate Orca-managed worktrees, folder contexts, terminals, repos, automations, artifacts, skill sharing, worktree comments, and the browser embedded inside the Orca app. Use when the user says "$orca-cli", "use orca cli", "Orca worktree", "child worktree", "cardStatus", "spawn codex/claude in a worktree", "read/wait/send Orca terminal", "terminal send", "full handoff", "handover", "give this to another agent", "another worktree", "Orca browser", "orca artifacts", "share HTML/Markdown", "public artifact link", "share skills", or "control the browser inside Orca". Prefer this over raw `git worktree`, ad hoc PTYs, Playwright, or Computer Use when the task touches Orca-managed state. Use Computer Use for browser windows, webviews, or desktop UI outside Orca's embedded browser.
Operate Orca-managed worktrees, folder contexts, terminals, repos, automations, artifacts, skill sharing, worktree comments, and Orca's embedded browser through the `orca` CLI. Use when the user says "$orca-cli", "Orca worktree", "child worktree", "spawn codex/claude in a worktree", "read/wait/send Orca terminal", "handoff" / "handover" / "give this to another agent", "Orca browser", "orca artifacts", or "share skills". Prefer it over raw git worktree, ad hoc PTYs, or Computer Use when Orca state is involved. Use Computer Use only when a visible window needs GUI control that a CLI, filesystem, or API cannot do.
When the user wants to design a multi-channel outbound sequence or cadence. Also use when the user says 'build me a sequence,' 'create a cadence,' 'how many touches,' 'what's the right sequence,' 'multi-channel outreach plan,' 'design my outbound,' 'how should I follow up,' 'SDR playbook,' 'outbound motion,' 'sales sequence,' 'follow-up strategy,' 'prospecting cadence,' or 'drip sequence.' For individual channel content, see cold-email, cold-call, linkedin-outreach, or direct-mail. For research before sequencing, see lead-research.
Write and edit styles in a project that uses Panda CSS. Use when creating, changing, or reviewing code that calls css(), cva(), sva(), patterns, recipes, tokens, conditions, or JSX style props, when a style renders a class name but no CSS, or when styles do not apply at all. Covers Panda v2 (2.x) and v1 (1.x). Do not use for designing the theme or deciding when something becomes a recipe or slot recipe (use panda-css-theming), or for upgrading v1 to v2 (use panda-css-migrate).
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.
