ChanMeng666

Visually verify a rendered SVG animation by opening it in chrome-devtools MCP, capturing keyframes, and scoring against the 5-point quality rubric. Invoke this immediately after `/svg-animate` writes an output, or whenever the user asks "is this animation good?", "how does it look?", "verify the motion", or "check the SVG visually". Always run before declaring an animation complete. Returns a ≤100-word verdict and never leaks screenshots into the calling thread.

Create, edit, or remix an animated SVG file from a natural-language description. Use this whenever the user asks you to make, build, design, generate, animate, or "draw something that moves" — mascots, logos, loaders, badges, icons, characters, brand marks, or any visual asset that should animate. Always prefer this skill over hand-writing SVG strings. Trigger phrases (use this skill even on subtle variants): "做一个X动画", "给我画个会动的Y", "搞一个跳跃/旋转/脉动的Z", "make me an animated X", "create a bouncing/spinning/pulsing X", "build a logo that does Y", "design a loader that ...", "I need an SVG mascot for ...". Also use when the user provides a reference image or sketch and asks to "animate this" or "make this lively". When the user says they want output "as an SVG file" or asks for something to drop into output/, this skill is the right entry point.

Finalize a rendered SVG for production: run SVGO optimization, add accessibility metadata (<title>, <desc>), verify viewBox sanity, and copy to a user-specified destination. Use when the user says "ship it", "optimize and export", "make it production-ready", "minify this SVG", "copy the final SVG to <path>", "export <slug> for the website", or asks for the byte-size after optimization. Always run after /svg-animate produces a passing verdict and before the SVG is committed or shared.

Add a new reusable motion / shape / filter primitive to lib/primitives/ with safety: JSDoc + sensible defaults + at least one consuming preset + passing snapshot tests + updated docs/lib-api.md + CLAUDE.md index. Use this only when /svg-animate has flagged a missing capability or when the user explicitly asks to "add a primitive", "extend the lib", "teach the system to <verb>", or "I need a new <motion / filter / shape> for ...". This is the only sanctioned path for growing lib/primitives/; do not edit those files directly.

Use whenever the user asks to install, configure, uninstall, snooze, mute, test, troubleshoot, or change settings for the echook audio notification system. Trigger phrases include "audio hooks", "audio notifications", "snooze audio", "mute claude", "claude is too loud", "test audio", "switch audio theme", "rate limit alerts", "audio webhook", "TTS", "text to speech", "notification mode", "audio only", "notification only", "debounce", "status line", "statusline", "context usage", "context window", "context monitor", "compact reminder", "uninstall audio", "audio status", "audio version", "install for cursor", "install for codex", "codex audio", "codex hooks", "cursor audio", "cursor hooks", and the slash command /audio-hooks. Also use when diagnosing why Claude Code, Cursor, or Codex is silent (or noisy) for the user, or when the user wants to monitor context window usage.

Generate professional PDF documents using the Typst typesetting system. Use when the user asks to create a PDF, write a document, generate a report, typeset a paper, build a presentation, or make a resume / CV / cover letter / letter / invoice / thesis / handout / poster — even when they do not say "Typst" explicitly. Especially trigger when the user mentions one-page CV, two-column resume, academic paper with bibliography, or any CJK (Chinese / Japanese / Korean) document.

Use when the user wants to create, edit, or inspect an architectural floor plan / building layout as code — e.g. "draw a 2-bedroom apartment", "add a bathroom to this plan", "make the bedroom 1 m wider", "lay out an office floor". ArchLang is a tiny text language that compiles a .arch file to a professional floor-plan drawing (SVG/PNG/PDF/DXF). Drive it entirely through the `arch` CLI; do not hand-render.

Capture polished screenshots and an optimized demo GIF/MP4 of a deployed web project and embed them into that project's GitHub README behind an idempotent SHOWCASE marker block. Use whenever the user asks to "add screenshots to a README", "record a demo of <project>", "showcase <project>", "update project images", "make a demo GIF", wants visitors to see what a project looks like without opening the live link, or wants any deployed project to look better on GitHub — even if they don't say "showcase". Works on any live URL; no install needed in the target repo. Generates assets locally and ALWAYS stops for user review before any commit; never pushes on its own.

Generate plain-Markdown "mirror" pages of a web project's important pages (pricing, docs, features, landing, long-form content) collected under a /llms/ folder — the neon.com/pricing.md pattern — so AI agents and crawlers parse them cleanly (GEO / AI SEO) and visitors can one-click "ask AI". Also builds an llms.txt index, injects rel="alternate" type="text/markdown" discovery hints in the page head, and can add "Open in ChatGPT/Claude" + "Copy as Markdown" buttons. Use this whenever the user wants .md mirrors, an llms.txt, an "ask AI / Open in ChatGPT" button, to make a site AI-readable / agent-friendly / LLM-friendly, to improve GEO or get cited by ChatGPT/Perplexity/Claude, or mentions agents getting confused by JS-heavy pages (sliders, interactive pricing, tabs, tables) — even if they don't say ".md mirror" explicitly.

Scaffold a complete set of professional, best-practice GitHub repository documentation for a project — README (agent-first), LICENSE, .github/FUNDING.yml, CODE_OF_CONDUCT.md, CONTRIBUTING.md, SECURITY.md, SUPPORT.md, GOVERNANCE.md, CHANGELOG.md, CITATION.cff, issue templates, PULL_REQUEST_TEMPLATE.md, AGENTS.md, and llms.txt. Use this whenever the user starts a new repo or wants to make an existing one look professional, polished, or "open-source ready" — e.g. "set up the docs for my new project", "add a README and license", "make this repo professional", "add community health files", "scaffold CONTRIBUTING / SECURITY / code of conduct", "prep this for GitHub", "add issue and PR templates", "create an AGENTS.md", "wire up GitHub Sponsors / FUNDING" — even when they don't name the specific files. This is Chan Meng's house style: it brands the README with her personal block, writes the README agent-first (so visitors point their AI agent at it), picks an appropriate license by project type, and copies canonical

Use when Chan Meng wants to write an English LinkedIn post for personal-brand marketing — either an "idea" post (sharing a framework/opinion) or a post-event "recap" (reflecting on an offline tech meetup, mentioning speakers/organizers/companies to grow her network). Drafts on-brand copy with emoji + hashtags, generates a Caldera-branded visual (Mermaid diagram, OpenAI image, or HTML card), and drafts it into the LinkedIn composer via claude-in-chrome for review — never auto-posts.

Use when turning a hand-drawn, sketched, or scanned logo into clean reproducible code — fitting smooth cubic Bézier curves to the real outline to produce an editable SVG, color/layout variants, an interactive side-by-side/difference review gallery, PNG/ICO/favicon exports, and optional AI showcase renders. For brand logos, monkey/mascot marks, wordmarks, avatars, or any artwork that should become a parametric, infinitely scalable, version-controlled asset instead of a flat image. This reconstructs an existing drawing; it does not generate brand-new logo designs from scratch.

Conduct a rigorous, two-lens peer review (editorial + technical-domain) of academic papers and produce a polished deliverable as Markdown, a Typst source, and/or a compiled PDF. Use this skill whenever the user asks to review, audit, critique, referee, proofread for rigor, or give peer-review feedback on a paper, manuscript, preprint, thesis chapter, or technical report — or to check a draft before submission, write a reviewer report or reviewer letter, or assess a paper's claims, statistics, citations, baselines, figures, or methodology — even if they never say the word "review". Especially relevant for ML / AI / RAG / IR / NLP / data-systems papers, where domain-specific methodological critique (missing baselines, gameable metrics, evaluation bugs) matters as much as editorial polish.

Generate cross-platform app promotional / marketing images (App Store & Google Play screenshots, iPad, Mac, desktop window, web landing hero, and social/Open-Graph cards) from the user's own app screenshots plus a few headlines and brand colors. Use when the user wants to create app store screenshots, promo graphics, marketing images, launch visuals, OG/social cards, or "宣传图" for an app — and wants exact platform sizes exported as PNG/JPG/WebP/AVIF and/or true vector SVG. Renders designed templates with a real browser (Playwright) and rasterizes vectors with resvg.