Skills
101,323 packaged capabilities and playbooks for agents. Install any of them with one command.
All
Deep performance optimization skill for Triton operators on Ascend NPU, dedicated to achieving the Triton operator performance improvements requested by users. Core techniques include but are not limited to Unified Buffer (UB) capacity planning, multi-token parallel processing, MTE/Vector pipeline parallelism, and mask optimization. Always trigger this skill when the user mentions the following: performance optimization of Vector-type Triton operators on Ascend NPU.
Risk-based vendor assessment framework for IT service providers, technology vendors, and third-party partners under DORA, NIS2, GDPR. Provides three-phase process (Initial Screening / Detailed Assessment / Final Evaluation), six-dimension risk scoring (Financial/Operational/Compliance/Security/Reputational/Strategic) with weighted matrices, full DORA Art. 28-30 contractual checklist, NIS2 Art. 21(2) security measures enumeration, GDPR Art. 28 documentation checks, red flags per dimension, trigger-based review criteria, and document templates. Use when: (1) Evaluating new vendors or technology providers, (2) Conducting critical ICT third-party due diligence under DORA, (3) Performing supply chain security assessment under NIS2, (4) Creating vendor onboarding documentation, (5) Establishing ongoing vendor monitoring, (6) Assessing concentration risk, or (7) Generating executive vendor risk reports.
Call POST /chat/completions on Venice. Covers the OpenAI-compatible request shape, Venice-only venice_parameters (web search, scraping, citations, E2EE, characters, thinking control, X search), anon_user_id, multimodal inputs (images/audio/video/files), tool calls, reasoning controls (reasoning_effort, reasoning.enabled), streaming, prompt caching, structured output, per-model caps, and model feature suffixes.
Find company and contact data. Turn your agent into a prospecting platform. Get contact information, roles, tech stacks, business events, website changes, and intent data. Build lead lists, research prospects, and identify talent. 150M+ companies, 800M+ professionals, 50+ data sources.
Use meitu-cli for comprehensive video quality restoration (deblurring/denoising/frame interpolation/brightening shadows), auto-adapting to portraits, concerts, anime, night scenes, and products. Trigger when the user mentions 画质修复, 视频修复, 变清晰, 老视频翻新, or one-click video enhancement without a resolution requirement.
Provides comprehensive guidance for Vuex 2.x state management in Vue 2 applications including state, mutations, actions, getters, modules, and plugins. Use when the user asks about Vuex for Vue 2, needs to manage state in Vue 2 applications, or implement Vuex patterns
Systematic analysis of Google Ads and Meta advertising spend to identify wasted budget. Produces actionable exclusion lists with thematic categorization and statistical validation. Use when onboarding new accounts, performing monthly hygiene, or preparing for budget optimization.
Reviews watchOS code for app lifecycle, complications (ClockKit/WidgetKit), WatchConnectivity, and performance constraints. Use when reviewing code with import WatchKit, WKExtension, WKApplicationDelegate, WCSession, or watchOS-specific patterns.
Generate a reusable design.md or DESIGN.md from a live website by deeply inspecting the site with `agent-browser` and `agent-browser eval`, then synthesizing its visual language, layout system, interaction patterns, and content style into a structured markdown design system. Use when given one or more website URLs and asked to analyze a site, reverse-engineer its design, extract its look and feel, write DESIGN.md, create a style guide, or capture a site's UI rules for later AI-assisted design or implementation. Prefer this skill when the goal is documentation rather than rebuilding the page in code. Always use `agent-browser` as the primary browser runtime; if it is missing, help the user install or expose it instead of switching platforms or browser stacks.
WebView traps across device classes: startup that freezes low-RAM phones, a killed renderer that takes the app with it, devices with no usable provider, AndroidView lifecycle and cookies. Use when a web screen freezes, vanishes or crashes only on some devices.
Graphics and media API reference for Windows app development (DirectX / WinRT media). Covers Direct2D (ID2D1Factory, ID2D1DeviceContext, ID2D1Bitmap1, ID2D1Effect, D2D1 Effects, ID2D1PathGeometry), DirectWrite (IDWriteFactory, IDWriteTextLayout, DWriteCore), Direct3D 11/12 (D3D11CreateDevice, D3D11On12, DXGI SwapChain, HLSL, D3D12 PipelineState, D3D12 DRED), Microsoft.UI.Composition (Compositor, Visual, ExpressionAnimation, InteractionTracker, MicaController), camera and screen capture (MediaCapture, MediaFrameReader, GraphicsCaptureSession, Windows Studio Effects, OpenCV), and audio and video playback (MediaPlayer, MediaPlaybackSession, AudioGraph, MediaTranscoder, AdaptiveMediaSource, HLS, DASH, CastingDevicePicker, AudioPlaybackConnection).
Use when an agent must act as a World Cup predictor for World Cup predictions, 2026 World Cup simulations, World Cup win/draw/loss probabilities, 90-minute match probabilities, explanations of champion or qualification probabilities, audited offline input checks, continuing a tournament based on completed results, or cautious China football lottery 3/1/0 reference lists
Use when a topic is best explained by YouTube videos — find and summarize YouTube videos for topics where visual explanation, demos, tutorials, talks, walkthroughs, or screen recordings are useful. Can run standalone, or export transcript-backed video source notes into notebooklm-mode vaults for grounded research.
Use when a framed and root-caused problem needs its assumptions made explicit before design or planning begins — explicit Assumptions, Unknowns, and a Validation plan. This should trigger when an issue's Assumption Analysis point of view needs evaluation, or when a maintainer directly asks to surface hidden assumptions and unknowns before committing to an approach. Part of Plinth Toolkit
Generate on-brand marketing images via Codex's built-in image_generation tool. Trigger when user asks to: (a) create marketing assets (ad / logo / slide / product-mockup / scene / lighting-transform / LinkedIn-or-social carousel) for a specific brand, (b) extract or build a brand profile (DESIGN.md) from URL / Tailwind config / tokens.json / Figma Variables / CSS custom props / description / existing brand asset, (c) maintain on-brand consistency across multiple image jobs for the same brand. Do NOT trigger for: UI code generation, frontend reference imagery (use imagegen-frontend-web instead), video generation, or general image editing without brand context. RUNTIME: Requires OpenAI `image_generation` tool (gpt-image-2 via Responses API). Native to Codex (CLI / App / web). Claude Code users need an MCP server that exposes image generation. If the tool is unavailable, this skill HARD-STOPS with remediation guidance — it does NOT fall back to HTML/CSS/Canvas rendering.
Inventory, classify, and pin every `uses:` reference in GitHub Actions workflows, composite actions, and reusable workflows to a verified commit SHA, then keep those pins fresh with Dependabot. Use when the user says "pin actions", "unpinned actions", "SHA pin", "supply chain for workflows", or invokes /actions-pinning. Do NOT use for permissions and injection hardening of workflow files (route to /oss-hygiene) or for CI run-time speed (route to /ci-optimize).
Creates and inspects Fabric Activator (Reflex) alerts: rules, their data sources, conditions, and the Teams or email action they fire, plus decoding an existing ReflexEntities definition. Choose it whenever someone wants to be told, or something done, when values cross a threshold. The pipeline delivering those values is eventstream-cli.
Compose custom agents from base traits, voices, and specializations for specialized perspectives. Create agent teams for parallel work. USE WHEN creating custom agents, spinning up agents, specialized agents, agent personalities, available traits, composing agents, agent profiles, spawning parallel agents, launching agents, agent teams, or swarms.
Create static, browser-based UI prototypes with a zero-build stack built from z-proto, HTM UI, Tailwind CSS v4, Preact/htm, and preact-iso. Use when asked for an HTML prototype, clickable browser mockup, interaction demo, or stakeholder flow that runs without a backend or build pipeline. Component, theme, and API implementation lives in the complementary htm-ui skill. For prototypes in a design tool (Paper, Figma, Pen.dev, Penpot), use agile-design.
Query Chinese and global financial data using the AKShare Python library. Use when asked to (1) fetch stock quotes, historical prices, or financial statements for A-shares/HK/US stocks, (2) query macroeconomic data like GDP/CPI/PMI, (3) get futures/options/bond/forex/fund data, (4) look up index data, or (5) retrieve alternative data like news sentiment. Covers 500+ data interfaces for stocks, futures, options, bonds, forex, funds, macro, indexes and alternative data.
Bring up a new Alchemy cloud provider (its own `@alchemy.run/<provider>` package under packages/<provider>) end to end — credentials, a distilled SDK, resources, bindings, live tests, and the patch loop that feeds live-test failures back into distilled — or extend an existing provider with a new service, or bring an existing provider up to the bar (gap audit, then waves). Use for "add a <X> provider", "create an alchemy provider for <X>", "port <X> to alchemy", "add <service> to the <X> provider", "bring the <X> provider up to par", or any work that needs a distilled SDK plus Alchemy resources tested against a live API.
Asciify attached or linked images, GIFs, video and 3D models: convert them to ASCII art with the asciify-engine command (text for a reply, ANSI for a terminal, SVG or animated SVG for a README), and build ASCII components and backgrounds for websites. Use when someone says asciify or asks for ASCII, text, braille or block art of a photo, logo, file or URL, a text banner, or ASCII in a terminal; and for interface components in characters (loading, status, toasts, tabs, AI states, charts, transitions, pages), animated backgrounds and heroes inked from the site's theme, live image, video and 3D fields on the GPU, artwork looks and art styles for wallpapers, posters and prints, hover and motion, type layered over media, export, rendering performance, and ASCII on servers and at the edge.
This skill should be used when users need to interact with AWS services via CLI. It covers all AWS services including EC2, ECS, EKS, Lambda, S3, RDS, DynamoDB, VPC, Route53, CloudFront, Bedrock, Support, Billing, and more. Supports querying, creating, modifying, deleting resources, monitoring, debugging, and cost analysis. Triggers on requests mentioning AWS, cloud resources, or specific AWS service names.
Use when the question is about money on a Bright Data account: current balance, credit left, what was charged, spend or cost history, credits or free tier used, cost per zone, product or dataset, or what a job will cost before it runs. Read-only, and it never buys or changes anything. Not scraping or extraction (use scrape), not how to drive the CLI itself (use brightdata-cli), not a refused or failing call (use agent-onboarding).
Build a distributable macOS DMG from an Xcode project via the bundled local helper — archive → export → package into one signed artifact. Resolves the project/workspace, scheme, team ID, and export options, regenerates the project first when `xcodegen` is detected, then runs `helpers/build-dmg.sh` from the project root instead of inlining xcodebuild. User-only workflow that touches code signing and provisioning: it never GUESSES the signing identity, never overwrites `ExportOptions.plist` without approval, and reports the built DMG path, version, and signing status. Use only when the user explicitly asks to build or package a DMG.
Guides the creation of career documents for academic advancement including research statements, teaching statements, diversity statements, CVs, and biosketches. Combines strategic positioning, narrative coherence, and institutional alignment with authentic representation of contributions and vision. Use when writing or reviewing faculty applications, fellowship statements, promotion packages, award applications, or when user mentions research statement, teaching philosophy, diversity statement, biosketch, academic CV, or career narrative.
Write nurb.app changelog entries from releases since the last published one. Filter the release's merged PRs to what users would actually notice, draft each entry in the site's voice, and insert the release into site/changelog.html. Use when the user says “changelog,” “update the changelog,” “announce this,” or after cutting a release.
Chaos engineering principles, controlled failure injection, resilience testing, and system recovery validation. Use when testing distributed systems, building confidence in fault tolerance, or validating disaster recovery.
Deep grill that interviews the user one question at a time AND verifies each answer against the codebase via background `answer-verifier-agent` calls. Catches factual conflicts, contradictions with prior decisions, and unfounded assumptions inline. Persists the session to `.chief/_grill/opened/NNNN-topic.md` so it survives context compaction. Use when the user wants a stress-tested grill — when stakes are high, when claims must be cross-checked against actual repo state, or when the topic spans many decisions. Heavier than `/grill-design`; prefer this when correctness matters more than speed.
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.
