Use when long-lived agents, project memories, system prompts, or playbooks are repeatedly summarized, rewritten, or optimized and risk losing domain detail, provenance, or execution learnings.
日本語の概要は準備中です。原文の説明を表示しています。
Build hyper-personalized, narrative-driven, single-file portfolio/pitch websites in the style of anass.pro. Use when the user asks for a "site like anass.pro", a one-recipient candidacy/pitch site, a behavioral-engineering landing page, a single-file storytelling portfolio with a pixel-art game hero, custom cursor, typewriter text, procedural Web Audio SFX, scroll-revealed scenes, or a dark purple narrative interface that addresses one named reader. Trigger on phrases like "personal pitch site", "one-page candidacy", "narrative portfolio", "scrollytelling pitch", "pixel-art hero", "interactive landing for X", or any request to recreate / clone / adapt anass.pro.
インストールする前に、エージェントに与えられる指示の中身を確認できます。
This skill captures the design system, interaction patterns, and authoring playbook behind anass.pro — a single-file, vanilla-JS, hyper-personalized pitch site that uses an interactive pixel-art game, procedural Web Audio SFX, scroll-revealed scenes and behavioral copywriting to address exactly one named reader.
Use it when the user wants to build something with the same DNA: a one-recipient narrative landing page that feels designed for them and only them.
Anass.pro is not a generic portfolio. It is a behavioral pitch built around four ideas:
<style> and <script>, vanilla JS classes, a Canvas 2D engine, and Web Audio API for procedural sound. No build, no React, no audio assets.If the user wants a generic personal portfolio, stop and switch to frontend-design. Use this skill only when the brief is a one-recipient pitch / candidacy / narrative landing page.
Task progress:
- [ ] 1. Confirm fit: is there ONE specific named recipient and a story to tell about them?
- [ ] 2. Gather inputs: recipient name + role, sender pitch, 2-4 specific facts about the recipient, the "ask" (CTA target), assets (logos), language/tone
- [ ] 3. Pick a scene sequence (use the 7-scene default, or adapt)
- [ ] 4. Lock the design tokens (see references/design-tokens.md) — pick palette + 4 fonts
- [ ] 5. Scaffold the single HTML file from templates/starter.html
- [ ] 6. Implement scenes top-to-bottom using references/section-recipes.md
- [ ] 7. Wire the signature interactions (references/animation-patterns.md)
- [ ] 8. Add procedural Web Audio SFX if appropriate (references/web-audio-sfx.md)
- [ ] 9. Add the pixel-art canvas hero if appropriate (references/pixel-game-canvas.md)
- [ ] 10. Verify reduced-motion + mobile breakpoints + a11y
Stop after step 2 and confirm the brief with the user before scaffolding. The whole point of the format is that it knows the reader; if you don't have specifics about the reader, the site collapses into another generic template.
Ask the user (briefly, one round of questions only):
tu. Match the recipient's culture. Default to confrontational + complicit, not corporate.If the user can't answer the "Who is the one reader?" or "What do they already do well?" questions concretely, don't build this. Push back: "this format only works when we can name specific things the reader has done. Otherwise it reads as fake intimacy. Want me to switch to a regular portfolio instead?"
Use this exact sequence unless the user has a strong reason to deviate. Each scene has a CSS hook, a data-s index for the dot-nav, and a psychological job.
| # | Section id | Title pattern | Job | Required elements |
|---|---|---|---|---|
| 0 | #game-hero | (no heading — the hero IS a game) | Capture: trigger pattern-recognition in the first 3 seconds | Pixel-art canvas, HUD with recipient's name, intro speech card, persistent CTA pill |
| 1 | #s-hook | "You did X. I do Y. Same principle." | Hook: state the analogy between the reader's work and the site itself | Brand bar, 3 typewritten lines, animated counter, italic aside |
| 2 | #s1 (Mirror) | "You used 500 X. I use one." | Mirror: prove you understand them by tabulating their move next to yours | 2-column compare table with row labels |
| 3 | #s2 (Method) | A quote from the recipient | Method: 3 expandable steps explaining the psychology of what just happened | 3-card flex-grow row, big italic ordinal numbers, closing line |
| 4 | #s3 (Reveal) | "What you just lived through is X." | Reveal: name the technique. Italic gradient title. | Centered, watermark logo glowing behind, body + pull quote |
| 5 | #s4 (Person) | "A. R." (sender) | Person: brief identity + stats + project grid | 2-col grid: stats left, bio + chips + project tiles right |
| 6 | #s5 (Close) | "Now you know what I do when I want something." | Close: the ask. One gradient CTA. Italic Fraunces sign-off. | Big condensed headline, body, single magnetic CTA, small meta line, italic sig |
The user is allowed to skip Mirror or Method, but never skip Hero, Reveal, or Close — they carry the format's identity.
Open references/design-tokens.md and pick (or accept the defaults):
cursor:none + custom dot+ring follower. Touch devices fall back to native cursor.Inject all tokens as :root CSS variables. Reference: references/design-tokens.md.
Start from templates/starter.html. It contains:
:root design tokensdata-s indicesGameEngine stub, a Typer class, the Splitter, the IntersectionObserver reveal, and the magnetic-CTA loopCopy it, then rename ids/classes only if you have a strong reason. The class names (.scene, .rv, .sw, .grad, .pu, .cs-arr) are referenced from multiple scripts and CSS rules — keep them.
For each scene, follow the recipe in references/section-recipes.md. The reference file gives you, per scene:
.rv, .rv.split, .d1/.d2/.d3 stagger, etc.)Two non-negotiables across all scenes:
<span class="pu">…</span> to highlight key purple words (the recipient's words, the specific numbers, the verbs of action).<span class="grad">…</span> to gradient-clip the most emotional phrase per scene — never more than one per scene. Overuse kills it.These six interactions are the format. Skip them and you have a dark purple landing page; include them and you have an anass.pro-style site. Implement in this order, each is independent:
.rv class fades up when in view; .rv.split splits text into per-word spans with --i index for staggered fade.Optional but high-impact:
All recipes in references/animation-patterns.md.
The site has zero audio files. Every sound is synthesized at runtime via AudioContext. This is fundamental to the "all-craft" feeling. Read references/web-audio-sfx.md for ready-to-paste recipes for:
key (typewriter click + thud) — fires on every typed characterclick (UI confirmation)ding (counter completion)tick (counter pulse)cta (CTA hover/click)spell (game projectile)explosion (game boss death — noise + lowpass sweep)monster (boss arrival — 4-layer: thud + sawtooth drone + tension sine + whoosh)jumpLand (character lands after a jump arc)Wrap them in a single SFX module with a global mute toggle, persisted to localStorage. Default to muted-but-armed: the toggle reads "Activer les sons" on first load — never autoplay.
The hero is the format's signature. If the project has time/budget, implement it. Reference: references/pixel-game-canvas.md. Key architecture:
<canvas id="game-canvas"> filling the viewportGameEngine class with update(now) / render() running on requestAnimationFramectx.fillRect)0=walk-in → 1/2/3=boss fights → 4=victory walk → 7=final revealIf the project is constrained, replace the game with a static animated illustration (canvas star field + parallax buildings, no characters/bosses). The narrative still works.
Non-negotiable checks before shipping:
@media (prefers-reduced-motion: reduce) collapses all transitions/animations to 0.01ms@media (hover:none) or width <= 768px)<div aria-live="polite">) announces phase transitions if you implement the gamealt; decorative images use aria-hidden="true" and empty alt<a href="…">, not a div| Trap | Fix |
|---|---|
| Generic copy ("I'm a passionate developer") | Quote the recipient. Reference their specific work. Use tu/intimate form. |
| Inter / Roboto / system fonts | Use the Barlow + Fraunces stack. Distinctive type carries the voice. |
| Gradient on every heading | Exactly one .grad per scene. The rest is .pu (purple ink) for emphasis. |
| Audio on autoplay | Default muted. Toggle is opt-in. |
| Heavy build pipeline | Single HTML file. No bundler. The constraint is the point. |
| Multiple CTAs | One CTA in the Close. The Hero CTA only appears after the user scrolls back to the top after reading. |
| Overdone parallax | 3 building layers max. More noise, less signal. |
| Skipping the Reveal scene | The Reveal is what reframes the whole experience as engineering. Without it the site reads as gimmick. |
references/design-tokens.md — Colors, type system, spacing scale, alternative palettesreferences/section-recipes.md — HTML/CSS skeletons for each of the 7 scenesreferences/animation-patterns.md — Cursor, reveal, typewriter, magnetic CTA, character buddy, counterreferences/web-audio-sfx.md — Procedural sound recipes (copy-paste functions)references/pixel-game-canvas.md — Game engine architecture, sprite drawing, parallax, bossesreferences/copywriting-playbook.md — Voice, tone, structure formulas for each scene's texttemplates/starter.html — A complete, working scaffold to copy fromRead each reference file only when you reach the corresponding step. They're all under 400 lines so a single read is cheap.
frontend-design insteadui-ux-pro-max or slidesまだレビューはありません。使ってみた感想をお寄せください。
概要と使いどころ
Use when long-lived agents, project memories, system prompts, or playbooks are repeatedly summarized, rewritten, or optimized and risk losing domain detail, provenance, or execution learnings.
日本語の概要は準備中です。原文の説明を表示しています。
Use this skill when the user explicitly asks to map, document, or onboard into an existing codebase. Trigger for prompts like "map this codebase", "document this architecture", "onboard me to this repo", or "create codebase docs". Do not trigger for routine feature implementation, bug fixes, or narrow code edits unless the user asks for repository-level discovery.
日本語の概要は準備中です。原文の説明を表示しています。
Adds educational comments to a code file so it becomes a learning resource, asking for the file (with a list of close matches) if none is given. Use when preparing code for learners, onboarding, or teaching.
日本語の概要は準備中です。原文の説明を表示しています。
Write, debug, and optimize Adobe Illustrator automation scripts using ExtendScript (JavaScript/JSX). Use when creating or modifying scripts that manipulate documents, layers, paths, text frames, colors, symbols, artboards, or any Illustrator DOM objects. Covers the complete JavaScript object model, coordinate system, measurement units, export workflows, and scripting best practices.
日本語の概要は準備中です。原文の説明を表示しています。
Patterns and techniques for adding governance, safety, and trust controls to AI agent systems. Use this skill when: - Building AI agents that call external tools (APIs, databases, file systems) - Implementing policy-based access controls for agent tool usage - Adding semantic intent classification to detect dangerous prompts - Creating trust scoring systems for multi-agent workflows - Building audit trails for agent actions and decisions - Enforcing rate limits, content filters, or tool restrictions on agents - Working with any agent framework (PydanticAI, CrewAI, OpenAI Agents, LangChain, AutoGen)
日本語の概要は準備中です。原文の説明を表示しています。
Check any AI agent codebase against the OWASP Agentic Security Initiative (ASI) Top 10 risks. Use this skill when: - Evaluating an agent system's security posture before production deployment - Running a compliance check against OWASP ASI 2026 standards - Mapping existing security controls to the 10 agentic risks - Generating a compliance report for security review or audit - Comparing agent framework security features against the standard - Any request like "is my agent OWASP compliant?", "check ASI compliance", or "agentic security audit"
日本語の概要は準備中です。原文の説明を表示しています。