Loading...
Loading...
Found 1,119 Skills
Hunting skill for xss vulnerabilities. Built from 174 public bug bounty reports. Use when hunting xss on any target. For markup injection that reflects raw HTML but does NOT execute JavaScript (no `<script>`/event-handler execution), see hunt-html-injection — escalate here once script execution is possible.
Manage Otter Drive artifacts with the otterdrive CLI. Use when an agent needs to authenticate with Otter Drive, select an organization, create or publish HTML, Markdown, CSV, TSV, Excel, image, or other static content, inspect or retrieve artifact files, manage immutable versions, update metadata, move artifacts between teams, open previews, or archive and restore artifacts.
Explains concepts, PRs, and designs for a specific reader as concise crash-course material without unnecessary verbosity, and verifies the diagrams and claims with tools. Build the reader persona—including what they already know, what they do not know, and how they read—from questions and public information, and write only the resulting delta. Create diagrams with vlmkit-anim against a fact sheet, re-run and cross-check code and output quoted in the body, and pass the HTML through vlmkit's validation gates. Use when the user says “explain this to me / to <person>”, “crash-course material”, “explanatory documentation”, “I want to understand this PR”, “explain it so I understand”, “I can't keep up with what the agent wrote”, or when a reviewer asks what a change does and the answer needs more than a paragraph. Also use it when the user wrote an explanation themselves and lost confidence in it.
Clone a website — two modes. SNAPSHOT = a faithful, self-contained offline copy of a page (HTML+CSS+JS+images in one file, exact bytes, SingleFile-style). RECONSTRUCT = AI reverse- engineer the page into a clean editable Next.js/Tailwind codebase (design tokens → component specs → parallel builders → visual-diff QA, the ai-website-cloner-template method). Use when the user wants to "clone this site", copy a page's design/UI + interactions, mirror a layout, save offline, or rebuild a site as editable code. Built-now-adapt-later: snapshot builtin works now; faithful live capture + the reconstruct pipeline are one-config engine adapters.
Crawl/scrape a website or single page into clean LLM-ready MARKDOWN. Use when the user wants to fetch a URL's content, scrape a site, extract an article/page as text, pull docs from the web into the wiki, or "crawl this link". Markdown (not raw HTML) = 5-10x fewer tokens. Built-now- adapt-later: works offline with no key via a builtin urllib→markdown backend; premium backends (Firecrawl/Crawl4AI/Jina) are a one-config adapter.
Tạo bộ test case / checklist UAT cho người dùng nghiệp vụ NON-IT (C&B, kế toán, vận hành) — 1 file HTML offline-proof gồm bảng kịch bản tuần tự có mã Ctrl+F, phạm vi check/không-check theo nguồn dữ liệu, step-by-step kèm ảnh màn hình thật đã blur PII, Đạt/Không đạt + xuất Excel/CSV/PDF. Trigger: "tạo test case cho non-IT", "checklist UAT cho nghiệp vụ", "kịch bản kiểm thử cho C&B", "/uat-nonit-testcase".
Design error responses. Use this when asked to implement, fix, or review 404 pages, error pages, Not Found handling, error handling, or API error responses, including choosing between 404, 410, and 301, avoiding soft 404s, serving HTML, Markdown, and application/problem+json according to Accept, implementing RFC 9457 Problem Details, defining cache policies, preventing internal information leakage, and validating responses with curl.
Start here for Yandex DataLens when the task has not yet settled on a tool, or when the question is about DataLens itself rather than about carrying something out in it: what DataLens is and what it can do; which installation is in play — Yandex Cloud, on-premise, or the internal Yandex Team one — and how they differ in hosts, auth, and capabilities; what a collection, workbook, connection, dataset, chart, or dashboard is and how they relate; and whether to reach for the MCP server, the Python SDK, the HTTP API, or the web UI. Use it to orient and to route to the skill that does the work. NOT for doing the work once the approach is chosen — that is `datalens-sdk` for Python and `datalens-html-pages` for standalone HTML reports — and not for interpreting what business metric values mean.
Create videos with code (Canvas / HTML-CSS-SVG / WebGL shaders / Three.js / Remotion / HyperFrames / Manim + code-synthesized sound), rendered frame by frame with Playwright and FFmpeg. It asks what the user wants to create (an idea, source material such as an article, script, product page or codebase, or a reference video), proposes three distinct video plans with shot-by-shot visual summaries, renders a key-frame still or a few-second preview for each, and, after the user confirms one, writes the full brief and builds the finished MP4. Use it whenever someone wants Claude to create, animate or render a video, motion graphic, animated explainer, product promo, kinetic-typography piece, pixel-art or retro animation, or UI motion demo, or wants to recreate the look of a reference video in code—including requests such as “make a video,” “make a video with code,” “create an animation,” “make a promotional video,” “help me recreate this video,” or “Opus, make a video.” Also use it when the user has AI-generated character clips (green screen) and wants code-built scenes around them. Do not use it for generating footage with video models, editing in Premiere/CapCut/AE, or subtitling an existing video.
Plan the opening blocking for every scene in an AI short drama: first identify objects in the scene image that people can stand beside or sit on and infer their positions on a top-down plane, then use the script to assign who is beside which object, on which side, whether sitting or standing, which direction they face, and whether extras are present. The program writes an opening blocking sentence for each scene, which the storyboard copies verbatim into that scene’s first segment. By default, deliver an interactive HTML tool report for dragging and adjusting positions; staging.json can be exported from the page. The JSON is also saved as working data, while Markdown is exported on demand. All eight quality gates are deterministically checked by script, including the number of recognized objects, whether they appear in the image, exact cast presence, whether everyone present is assigned to a recognized object, whether people overlap on the same side, and whether extras’ locations are specified. It consumes script.json from novel-script and art.json from novel-art to prefill a skeleton. Zero dependencies and zero API keys. Use when asked about staging, opening staging, ensemble staging, who sits where, seating arrangements, movement blocking, blocking, or staging for an AI short drama.
UI/UX design intelligence for web and mobile. Includes 50+ styles, 161 color palettes, 57 font pairings, 161 product types, 99 UX guidelines, and 25 chart types across 10 stacks (React, Next.js, Vue, Svelte, SwiftUI, React Native, Flutter, Tailwind, shadcn/ui, and HTML/CSS). Actions: plan, build, create, design, implement, review, fix, improve, optimize, enhance, refactor, and check UI/UX code. Projects: website, landing page, dashboard, admin panel, e-commerce, SaaS, portfolio, blog, and mobile app. Elements: button, modal, navbar, sidebar, card, table, form, and chart. Styles: glassmorphism, claymorphism, minimalism, brutalism, neumorphism, bento grid, dark mode, responsive, skeuomorphism, and flat design. Topics: color systems, accessibility, animation, layout, typography, font pairing, spacing, interaction states, shadow, and gradient. Integrations: shadcn/ui MCP for component search and examples.
Toolkit for styling artifacts with a theme. These artifacts can be slides, docs, reportings, HTML landing pages, etc. There are 10 pre-set themes with colors/fonts that you can apply to any artifact that has been creating, or can generate a new theme on-the-fly.