Skills
98,284 packaged capabilities and playbooks for agents. Install any of them with one command.
Development
Add Anthropic Claude models (Opus, Sonnet, Haiku) to Microsoft Teams.ai applications using @youdotcom-oss/teams-anthropic. Optionally integrate You.com MCP server for web search and content extraction. - MANDATORY TRIGGERS: teams-anthropic, @youdotcom-oss/teams-anthropic, Microsoft Teams.ai, Teams AI, Anthropic Claude, Teams MCP, Teams bot - Use when: building Microsoft Teams bots with Claude, integrating Anthropic with Teams.ai, adding MCP tools to Teams applications
Develop, debug, and manage Temporal applications across Python, TypeScript, Go, Java, .NET, Ruby, and Rust. Use when the user is building workflows, activities, workers, or background job queues with a Temporal SDK, debugging issues like non-determinism errors, stuck workflows, or activity retries, using Temporal CLI, Temporal Server, or Temporal Cloud, or working with durable execution concepts like signals, queries, heartbeats, versioning, continue-as-new, child workflows, or saga patterns. Also use when the user mentions "run a Temporal workflow from the CLI", "start a dev server", "run temporal server start-dev", "temporal workflow start", "temporal workflow execute", "temporal workflow signal", "temporal workflow query", "temporal workflow update".
Best practices for integrating Terra API – the unified health & fitness data API for 500+ wearables (Garmin, Fitbit, Oura, Whoop, Apple Health, Strava, Dexcom). Use when building with Terra API or tryterra.co, handling terra-signature (X-Terra-Signature) webhooks, storing wearable health data (activity, sleep, daily, body, nutrition, menstruation, hormone), managing device connections, or merging data across multiple devices.
Integrate TheSys C1 Generative UI API to stream interactive React components (forms, charts, tables) from LLM responses. Supports Vite+React, Next.js, and Cloudflare Workers with OpenAI, Anthropic Claude, and Workers AI. Use when building conversational UIs, AI assistants with rich interactions, or troubleshooting empty responses, theme application failures, streaming issues, or tool calling errors.
Use this skill when a terminal UI freezes, hangs, shows a blank screen, corrupts terminal state, leaves raw mode stuck, stops responding to input, garbles output, breaks colors/cursor state, fails over SSH/tmux/screen, behaves differently on Windows/ConPTY, or needs deep root-cause analysis for subtle terminal failure modes. PROACTIVELY activate for: TUI freeze, terminal hang, blank TUI, raw mode stuck, stdin paused, no input events, alternate screen stuck, mouse mode stuck, bracketed paste markers, garbled terminal output, corrupted cursor/colors, partial renders, resize races, terminal state cleanup bugs, npm run tui froze, and debugging terminal gotchas. Provides: freeze/hang diagnosis, terminal recovery, startup and cleanup lifecycle checks, platform triage, and fix patterns.
tmux for AI agents. Run dev servers and TUIs in named background sessions that agents can read, wait on, snapshot, and type into. Replaces tmux with reactive waiting instead of blind `sleep`. Projects wrap their dev script with tuistory (`"dev": "tuistory -- next dev"`) so agents get a background session and humans get auto-attached. Use tuistory when you need to: - Run background dev servers or long-lived processes - Control interactive CLIs and TUIs (type, press keys, click, wait, snapshot) - Write Playwright-style tests for terminal apps **CLI** (`tuistory`) for background sessions. **JS/TS API** (`launchTerminal`) for programmatic control and tests.
Adds, removes, upgrades, and discovers Unity (UPM) packages from outside the Editor, with the Pipeline package commands first and a C# script as the fallback. Use for package installs, verifying package IDs and versions, or choosing which packages a game needs.
Use the jev CLI when a script, pipeline, CI job, or agent workflow needs a fast typed semantic judgment instead of generated text: routing or classifying text to predefined labels, yes/no gating with calibrated probability, rating along ordered levels, ranking candidate texts, or checking a condition that would otherwise require a fragile prompt-and-parse step. Also use whenever the user mentions Jev, TypeSafe, System One, calibrated judgments, or semantic gates in automation. Do not use it for text generation, multi-step reasoning, or open-ended questions.
Use when you need to run interactive CLI tools (vim, git rebase -i, Python REPL, etc.) that require real-time input/output - provides tmux-based approach for controlling interactive sessions through detached sessions and send-keys
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.
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.
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.
Use when something moves in a three.js or React Three Fiber scene: rigs, clips, blending, retargeting, morph targets, crowds. NOT for scroll or UI motion (sheleg-design), rendered video (hyperframes) or Unity/Unreal. Opt out: no web3d, без web3d. Triggers - "AnimationMixer", "анимируй персонажа".
Use when building or fixing a realtime 3D web scene in three.js or React Three Fiber: WebGPU with WebGL2 fallback, TSL. NOT for page heroes (sheleg-design), assets (web3d-assets), animation (web3d-animation) or Quest (xr-dev). Opt out: no web3d, без web3d. Triggers - "3D-сцена тормозит".
WordPress Playground review and development guidance. Use when reviewing Blueprints, Playground JSON setup flows, `@wp-playground/cli` usage, embedded Playground demos, reproducible bug repros, docs examples, or when user mentions "WordPress Playground", "Blueprint", "playground.json", "run-blueprint", "build-snapshot", "embed Playground", or "zero-setup repro". Helps review Blueprint design, local Playground workflows, demo reliability, and reproducible WordPress environments.
Use when you want to generate Architecture Decision Records from this session. Triggers on \"write ADRs\", \"document our decisions\", \"create decision records\", \"record the choices we made\". Also useful after design discussions where decisions were reached but not documented. Does NOT extract decisions alone (use adr-decision-extraction) or provide MADR template (use adr-writing). Orchestrates the full workflow: subagent extraction, user confirmation, parallel generation, and verification.
Decompose a spec's PRD/TechSpec into a dependency-ordered task graph — vertical-slice task files plus a machine-parseable _tasks.md DAG manifest under docs/specs/<slug>/, including the authored QA gate decision, decided autonomously from the Spec and escalated only when authority, architecture, or blast radius is at stake.
Operate a Seeed XIAO ESP32S3 board with arduino-cli on Windows: install the core, compile, upload, read serial output, and run verified examples (blink, serial, touch, ADC, PWM, WiFi scan/AP, BLE, deep sleep). Use this skill WHENEVER the task involves an ESP32S3 / XIAO board, arduino-cli, flashing a sketch, reading the serial monitor, COM ports, or any embedded/Arduino work — even for a "simple" LED change. It contains verified commands and known pitfalls (DTR reset, deep-sleep USB drop, PSRAM flag) that prevent wasted debugging time.
ZenTao CLI shared foundation: config initialization, token-based authentication, config precedence, permission error handling, update checks, security rules. Triggers when the user says: 'configure zentao', 'init config', 'config init', 'auth', 'login zentao', 'ZenTao login', 'token', 'view config', 'config show', 'config set', 'environment variables', 'ZENTAO_URL', 'ZENTAO_TOKEN', 'insufficient permission', 'auth status'.
Zod 4 — TypeScript-first schema validation with static type inference. Use when writing Zod schemas, validating data, defining types with Zod, parsing input, creating form validation schemas, defining API request/response schemas, working with z.object, z.string, z.number, z.enum, z.array, z.union, z.discriminatedUnion, z.file, z.jwt, z.email, z.uuid, z.url, z.codec, z.toJSONSchema, z.fromJSONSchema, z.int, z.stringbool, z.templateLiteral, z.record, z.partialRecord, or any other Zod API. Also use when migrating from Zod 3 to Zod 4, or when the user's package.json shows zod@^4. CRITICAL: Always use Zod 4 APIs. Never use deprecated Zod 3 patterns unless user explicitly requests Zod 3 compatibility.
Build a multi-agent AG2 network — load this whenever two or more `Agent`s need to interact. The network is the standard multi-agent pattern in `ag2`. Covers `Hub.open`, `LocalLink`, `HubClient.register`, `Passport`, `Resume`, channel lifecycle (PENDING → ACTIVE → CLOSING → CLOSED), the two 2-party channel adapters (`consulting` for strict 1Q1R and `conversation` for free-form), the `Envelope` wire format, audience routing, `wait_for_channel_event`, WAL replay via `hub.read_wal`, and the five channel-close routes. Entry point — routes to `ag2-network-discussion` (N-party round-robin), `ag2-network-workflow` (declarative orchestration / GroupChat migration), `ag2-network-governance` (rules / expectations / audit), or `ag2-network-tools-and-views` (custom handlers / peer discovery) for deeper needs.
Shape what an AG2 network agent perceives and which actions its LLM can take. Covers the five auto-injected LLM-facing tools that ship via `NetworkPlugin` (`delegate`, `peers`, `channels`, `tasks`, `context`) plus the adapter-owned `say`; replacing the default handler with `agent_client.on_envelope(callback)` (gateways, headless workers); the `ViewPolicy` Protocol with the built-in `FullTranscript`, `WindowedSummary(recent_n=N)`, and `NamedWindowedSummary(recent_n=N)` views plus how to write a custom view; peer discovery via skill markdown (`skill_md=`, `parse_skill_frontmatter`, `hub.set_skill`); the `Envelope` wire format with the `EV_*` event taxonomy (`EV_TEXT`, `EV_PACKET`, `EV_CHANNEL_*`, `EV_EXPECTATION_VIOLATED`, `EV_TASK_CANCEL_REQUEST`, `EV_TASK_CANCELLED`), `audience` and `visible_to` semantics, `Priority`, `causation_id`, and how to send raw envelopes with custom event types via `agent_client.send_envelope(...)`. Use when the user wants to customise the LLM's network surface, write a custom envelope handler, build a gateway / headless worker, or wire peer discovery.
Build an AG2 network `workflow` channel — the orchestrated N-party adapter driven by a declarative `TransitionGraph`. Use when the user needs conditional handoffs, multi-step pipelines, context-aware routing, feedback loops, or is migrating from the classic `GroupChat` + `Agent.handoffs` pattern. Covers `TransitionGraph`; factories `TransitionGraph.sequence(...)` / `.round_robin(...)`; built-in targets (`AgentTarget`, `RoundRobinTarget`, `StayTarget`, `RevertToInitiatorTarget`, `TerminateTarget`); built-in conditions (`Always`, `FromSpeaker`, `ToolCalled`, `ContextEquals`); the typed `Handoff` return for dynamic routing; channel-scoped context via `set_context`; `register_target` / `register_condition` for custom subclasses; the packet execution model and idempotent-tool requirement; eight cookbook patterns (pipeline, hierarchical, escalation, feedback loop, triage, and more); and side-by-side migration from classic `GroupChat`. Load this after `ag2-network-quickstart`.
Framework for long-running AI agents across multiple context windows. USE WHEN user types /agent-harness or mentions sprint planning, feature breakdown, multi-session projects. Subcommands: init, sprint, code, 996, status, archive, force-archive.
Use when the user needs Agently TaskDAG / Dynamic Task, model-generated or app-submitted DAG planning, TaskDAG validation, TaskDAGResolver handlers, TaskDAGExecutor execution, or the Agently.create_dynamic_task compatibility facade. TaskDAG is the DAG foundation capability; Dynamic Task is the convenience facade over it and uses TriggerFlow as the execution substrate.
Design and integration guidance for adding a reusable AI Agent chat interface to an existing web application. Use when adding an embedded AI assistant, chat view, multimodal input, speech-to-text, session history, long-task execution UI, Agent activity states, provider/model settings, or Tool / Function Calling support. The Agent is a supporting interface for the existing application rather than the application's primary UI.
Answer questions about the AI SDK and help build AI-powered features. Use when developers: (1) Ask about AI SDK functions like generateText, streamText, ToolLoopAgent, embed, or tools, (2) Want to build AI agents, chatbots, RAG systems, or text generation features, (3) Have questions about AI providers (OpenAI, Anthropic, Google, etc.), streaming, tool calling, structured output, or embeddings, (4) Use React hooks like useChat or useCompletion. Triggers on: "AI SDK", "Vercel AI SDK", "generateText", "streamText", "add AI to my app", "build an agent", "tool calling", "structured output", "useChat".
Switch AI providers or models without breaking things. Use when you want to switch from OpenAI to Anthropic, try a cheaper model, stop depending on one vendor, compare models side-by-side, a model update broke your outputs, you need vendor diversification, or you want to migrate to a local model. Also use when your prompt broke after a model update, prompts that work for GPT-4 do not work for Claude or Llama, or you need to do a model migration. Covers DSPy model portability with provider config, re-optimization, model comparison, and multi-model pipelines. Also used for migrate from OpenAI to Anthropic, GPT to Claude migration, try Llama instead of GPT, model comparison framework, multi-provider AI setup, avoid vendor lock-in for AI, prompts break when switching models, model-agnostic AI code.
Andrej Karpathy's thinking frameworks and way of expressing ideas. Systematically distilled from 20+ blog posts, 16 in-depth interviews and 100+ X posts, extracting 6 core mental models, 8 decision heuristics, complete Chinese output adaptation and a quick reference of classic phrasings. Use as a thinking advisor, analyzing the reliability of AI technology, learning methods, industry trends and product design from Karpathy's perspective. Activates when the user mentions "from Karpathy's perspective", "what would Karpathy think", "Karpathy" or "karpathy mode". Also applies to: Software 2.0/3.0 discussions, vibe coding topics, neural network training, AI hype assessment and the limits of LLM capabilities. It can also trigger when the user merely says "from an engineering realism angle", "march of nines", "building is understanding" or "jagged intelligence". Do not trigger when the user is simply asking ordinary AI-related questions — activate only when they explicitly want Karpathy-style thinking frameworks.
UIKit iOS/iPadOS UI framework. UIViewController, UINavigationController, UITabBarController, UISplitViewController, UIView, UILabel, UIButton, UITextField, UIScrollView, UIStackView, UITableView, UICollectionView, diffable data source, NSDiffableDataSourceSnapshot, UICollectionViewCompositionalLayout, Auto Layout, NSLayoutConstraint, NSLayoutAnchor, UILayoutGuide, intrinsicContentSize. UIArrangementViewController, UISplitArrangement, UIOverlayArrangement, UIView.ReservedRegion, UIHingeInteraction, vertical bar, iPhone Duo adaptive layout (iOS 27.1+), migration to the scene-based life cycle, UISceneAccessory. A separate API from SwiftUI ArrangementView.
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.