Render a static iOS QWERTY keyboard as inline HTML+CSS sized for the 750-wide 9:16 stage. Includes the 3-suggestion bar, alpha keys, shift/backspace, 123/emoji/space/return row, and the bottom globe+mic strip. No animation logic — slide up/down is the molecule's job (CSS transform on the .keyboard root).
Install gooseworks: https://gooseworks.ai/creative-agent/install.md # then, in Claude Code, Cursor, or Codex: /gooseworks use the render-ios-keyboard skill
Superseded: the video kit now does this with the phone-chat part, version 1.2.0, in the parts folder of this repository. Every phone-chat style is a style file that draws the same screen frame by frame from the plan's scenes, with the original sounds, in the kit's browser. This atom stays, unchanged in behaviour, for skills outside the kit until they move; its scripts still run.
Provide a drop-in iOS QWERTY keyboard fragment that pairs with messaging atoms (create-chatgpt-mockup, create-imessage-mockup) when a video needs to show someone typing. Static only: the keyboard sits at the bottom of the screen, the molecule's recorder slides it up/down with transform: translateY(...).
const { renderKeyboardHTML, renderKeyboardCSS } = require('./generate');
const html = renderKeyboardHTML({
suggestions: ['"He"', 'Hey', 'Heating'], // optional, default ['I', 'The', "I'm"]
layout: 'qwerty-lower', // 'qwerty-lower' | 'qwerty-upper' | 'numbers'
});The atom returns an HTML fragment (one <div class="ios-keyboard">…</div>) plus the matching CSS. The CSS uses scoped class names so it won't collide with the rest of the page.
const { renderKeyboardHTML, renderKeyboardCSS } = require('./generate'); — no build step, no dependencies beyond Node's fs/path.renderKeyboardHTML reads opts.layout and looks it up in ROWS: qwerty-lower (default), qwerty-upper, or numbers. An unknown layout silently falls back to qwerty-lower. The molecule decides which state to show; this atom never switches on its own.opts.suggestions is an array of up to three strings rendered into the suggestion bar; each value is run through escapeHTML so quotes/angle brackets are safe. Omit it to get the default ['I', 'The', "I'm"].<div class="ios-keyboard" data-layout="…"> containing, in order: the .kb-suggestions bar, three letter rows (row-1 10 keys, row-2 9 keys padded with two 18px spacers, row-3 = shift/#+= modifier + 7 keys + backspace), the row-bottom (123/ABC switch, emoji, wide space, return), and the .kb-system-row globe + mic strip.renderKeyboardCSS() (which reads templates/keyboard.css from disk) and place it in a <style> block on the page. The molecule mounts the fragment at the bottom of the stage next to .composer-wrap.transform. The owning molecule slides the keyboard in/out via transform: translateY(...) on the .ios-keyboard root (the CSS sets will-change: transform for this).render.js wraps the fragment + CSS in a 750×1334 white .stage HTML page and writes it to --out for eyeballing padding/colors.node render.js --out /tmp/kb.html --suggestions 'I,The,I'\''m'Produces a standalone HTML page with just the keyboard, useful for tweaking padding/colors.
| Output | Shape |
|---|---|
renderKeyboardHTML(opts) | HTML string (one root <div class="ios-keyboard">) |
renderKeyboardCSS() | CSS string (scoped to .ios-keyboard *) |
The molecule injects these into the chatgpt-mockup page next to .composer-wrap.
| File | Purpose |
|---|---|
generate.js | Returns {renderKeyboardHTML, renderKeyboardCSS} |
render.js | Standalone CLI for visual review |
templates/keyboard.css | All keyboard styling |
layout but are visual-only — the molecule decides which to show.renderKeyboardHTML() returns a single root element starting with <div class="ios-keyboard" and carrying the matching data-layout attribute (qwerty-lower | qwerty-upper | numbers)..kb-suggestions bar, three letter rows (.row-1, .row-2, .row-3), one .row-bottom, and one .kb-system-row — in that DOM order.suggestions array (default ['I', 'The', "I'm"]); the bar renders one .kb-suggestion span per entry with vertical dividers between adjacent pills.qwerty-lower renders lowercase letters and a shift glyph; qwerty-upper renders uppercase letters; numbers renders the digit/symbol rows with the #+= modifier and an ABC switch key instead of 123.'<b>' or '"He"' yields <b> / "He", never raw markup.renderKeyboardCSS() returns the full contents of templates/keyboard.css, scoped under .ios-keyboard, and the fragment carries no inline style except the two 18px row-2 spacer divs.123/emoji/return modifier keys are visibly wider than letter keys; space is one wide white pill; the globe sits bottom-left and mic bottom-right in the system strip.<script> and no JS event handlers — so a recorder can slide it with a CSS transform without re-rendering.layout value → ROWS[layout] is undefined, so the renderer silently falls back to qwerty-lower. Symptom: you asked for uppercase or numbers and got lowercase letters. Fix: pass exactly 'qwerty-lower', 'qwerty-upper', or 'numbers'..kb-suggestion (they flex: 1 1 0), so a long array squashes the pills and breaks the iOS-accurate 3-up layout. Keep suggestions to three or fewer; long strings are clipped with text-overflow: ellipsis.renderKeyboardHTML() without also emitting renderKeyboardCSS() produces an unstyled stack of divs (no rounded keys, no gray tray). Always include both on the page.templates/keyboard.css missing or unreadable → renderKeyboardCSS() calls fs.readFileSync and will throw ENOENT. Run from a checkout where templates/keyboard.css sits next to generate.js.render.js run without --out → the CLI prints the usage line and exits with code 1; nothing is written. Always pass --out <path>.Research supported advertising promises from a specific buyer's context, current product facts, customer language and observed ads. Turn them into static, video or copy ideas; match video ideas to actual recipes and save a product-scoped angle bank for write-video-ad-script. Product-led discovery can use verified facts without reviews. Reference persistence and reach are observations, not proof of performance.
Analyze ad campaign performance data (Google, Meta, LinkedIn) to identify what's working, what's wasting budget, and specific cut/scale/test recommendations. Runs statistical analysis, funnel diagnostics, and multi-channel budget reallocation with specific dollar-amount shift recommendations and scenario modeling.
For paid lead-gen and participant-recruitment ads, replaces vanity CPA with true CAC per qualified lead by joining ad-platform data with downstream funnel events, surfaces tracking gaps, and classifies every creative into Scale / Keep / Investigate / Cut.