Index of every page on this site: llms.txt
All skills
Use this Skill: https://skilld.dev/gh/emilkowalski/skills/ask-emil

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

SKILL.md

≈23 tokens for metadata: the name and description. ≈1.2k when used: this file.

How this Skill uses context

Before choosing a Skill, your Agent reads its name and description. All available Skills share that space.

This Skill turns off automatic use in Claude Code. Its description stays out of the automatic Skill list by default.

Example settings and source

The example uses a 200k-token context and default Claude Code settings. The count includes the name, description, separators, and when_to_use when present. Codex also counts local file paths.

Skit's source and limits: Codex 0.160.1, Claude Code 2.1.292.

Ask Emil

Before stating what a skill does or recommending one, read that skill's SKILL.md: the summaries here are for orientation only. Then recommend one skill (or one short sequence) for the user's situation, say why in a sentence, and name the skill they'd reach for next. Don't list the whole map back at them.

A flow is a path through the skills. Most UI work travels one main flow, from idea to shipped. Some skills are upkeep you run on a codebase that already exists, some are references that sit underneath everything, and a few stand alone.

The main flow: idea → ship

  1. Not sure what it should look or feel like? Detour through /prototype. It builds several genuinely different versions behind a live picker, then promotes the one you pick into the real code. Reach for it whenever a design question is hard to settle in words.

  2. Build it. Pick the builder by what's moving:

    • /pick-ui-library first if the task needs a dependency (dropdown, dialog, toast, drawer, command menu, drag and drop, charts). Don't let the agent hand-roll a toast.
    • /animate for one animation on the web: whether to animate at all, then tool, properties, curve, duration, interruption, exit.
    • /choreograph when several elements move together: staggers, hero entrances, view swaps, an item leaving while siblings close the gap.
    • /gestures when the user's finger drives it: drag, swipe, sheets, snap points, swipe-to-dismiss.
    • /animate-expo for any of the above in React Native or Expo.
    • /ask-sonner when the work is toasts with Sonner.
  3. Try to break it. Before calling it done:

    • /break-ui feeds it worst-case data (long names, empty lists, huge counts, emoji, non-Latin text) behind a toggle and reports what broke.
    • /review-animations reviews the motion in the diff against a strict bar. Default to flagging; approval is earned.
    • /mobile-native if it runs on a phone as a web app: sticky hover, the 100vh bug, inputs that zoom, safe areas.
    • /animation-performance if anything drops frames.
  4. Learn from the session. /learn-taste closes the loop. Run it at the end, in the same session, before you clear: it collects every design correction you made (including edits you made yourself) and turns each into a rule with its why, so the next agent doesn't repeat the mistake.

Upkeep: a codebase that already exists

Not feature work. Run these when you have a spare moment, or when the app "feels off".

  • The motion needs work across the app → /improve-animations. Read-only: it audits every animation and writes prioritized, self-contained plans another agent (or a cheaper model) can execute.
  • It feels static → /find-animation-opportunities. Read-only: it proposes motion with exact values, and rejects everything that shouldn't move. Build the ones you accept with /animate.
  • It feels slow or over-animated → /remove-animations. Edits code, but only to subtract: motion on keyboard actions, high-frequency surfaces, decoration with no purpose.

These three cover different questions. If the user isn't sure which, ask whether the problem is bad motion, missing motion, or too much motion.

References underneath

Model-invoked skills the others lean on. Reach for them directly when the knowledge, not a process, is what's missing.

  • /emil-design-eng: the core philosophy for UI polish, component interactions, and the invisible details. The fallback when no narrower skill fits.
  • /apple-design: Apple's principles for interface design and fluid, physical motion, translated for the web.
  • /animation-vocabulary: when the problem is the word. "What's it called when…" turns a vague description into the exact term to prompt with.

Standalone

  • /write-swift: modern Swift, Swift 6 concurrency, and native iOS code. Off the web flow entirely.

Common situations

You say Reach for
"Animate this" /animate
"These should come in one after another" /choreograph
"It doesn't follow my finger" /gestures
"It stutters" / "it's janky" /animation-performance
"Does this animation feel right?" /review-animations
"Make the app feel better" /improve-animations
"What could animate here?" /find-animation-opportunities
"Too much is moving" /remove-animations
"Give me a few options" /prototype
"Which library should I use?" /pick-ui-library
"Works in Chrome, feels wrong on my phone" /mobile-native
"Try the worst case" /break-ui
"What's that bouncy thing called?" /animation-vocabulary
"Stop making that mistake" /learn-taste

Source: SKILL.md on GitHub

No rule matched.

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No alerts5d3 checks · Risk SAFE
  • Gen Agent Trust Hub5d

    The skill acts as a documentation router to help users find other animation and UI development skills. It contains no executable code or external connections. A low-risk surface for indirect prompt injection exists as it processes user input to recommend other tools.

  • Socket5d

    No alerts

  • Snyk5d

    Risk: LOW · No issues

Signed by skilld at 64eb3de. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 6 days ago.

Activeupdated 6 days ago
disable-model-invocation
true

README badge

README badge for emilkowalski/skills/ask-emil