Build multimodal products
Skill hiteshbandhu/skills-i-use/skills/ai-engineer-talks/build-multimodal-products
Runs workflows for multimodal AI products — realtime voice latency, tiny VLMs, SSM streams, unbounded UX design, dataset hierarchy, AI education. Use when the user builds voice bots, edge vision, multimodal UX, or says "multimodal product", "voice latency", "Moondream".From its SKILL.md
npx -y skills add hiteshbandhu/skills-i-use --skill build-multimodal-productsAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
- runs commandsInstructs the agent to run 3 commands, including `cp -r skills/build-multimodal-products ~/.claude/skills/` and 2 more.
SKILL.md
2.2 KB, 491 tokens by cl100k_base, as published. Nobody here has run it
Build multimodal products
Action playbook from six AI Engineer / World's Fair talks. Do not summarize talks — pick a workflow and execute it.
Supporting files (read when needed):
- workflows.md — workflows A–F (steps, deliverables, stop conditions)
- source-index.md — src-NNN → talk learnings in ingest-into-skills
Optional: {SKILL_OUTPUT_DIR}/build-multimodal-products/
Step 0 — Pick workflow
Use the decision tree below. Open the matching section in workflows.md.
What is the user trying to do?
├─ Ship realtime voice agent → A
├─ Deploy edge/tiny vision (Moondream-class) → B
├─ Long multimodal streams (SSM) → C
├─ Design unbounded multimodal UX → D
├─ Build training/eval datasets → E
└─ EdTech multimodal literacy → F
Stop summarizing once a workflow is identified — run its checklist.
Install
cp -r skills/build-multimodal-products ~/.claude/skills/
cp -r skills/build-multimodal-products ~/.cursor/skills/
cp -r skills/build-multimodal-products ~/.codex/skills/
From skills-i-use or ingest-into-skills (playlists/multimodality-aie-world-s-fair-2024/).
Cross-cutting rules
| Rule | Source |
|---|---|
| Measure time-to-first-audio | [src-001 @ 9:28] |
| Show model limits in UX | [src-004 @ 14:11] |
| Dataset legal/diversity before scale | [src-005 @ 1:38] |
Disputed steps: see source-index.md. Name workflow A–F; save artifacts to ./skill-outputs/build-multimodal-products/ when requested; do not auto-commit.
Invocation examples
@build-multimodal-products design voice agent under 500ms
multimodal UX without form fields
What ships with it: 3 files
3.4 KB alongside SKILL.md
- README.md920 B
- source-index.md1.5 KB
- workflows.md944 B
Gives 0 of the 12 instructions most design frontend skills give in 491 tokens
Counted across 1,179 of the 2,086 authors here whose files we hold, read 2026-09-06
- Commit to a bold aesthetic directionin 31 of 1179, across 24 files
- Prefer component composition over inheritancein 28 of 1179, across 14 files
- Animate only transform and opacity propertiesin 27 of 1179, across 22 files
- Memoize expensive computations with useMemoin 26 of 1179, across 13 files
- Use semantic HTML elementsin 24 of 1179, across 23 files
- Virtualize long lists for performancein 21 of 1179, across 10 files
- Use CSS variables for design tokensin 20 of 1179, across 14 files
- Implement loading, empty, and error statesin 20 of 1179
- Lazy load heavy components with Suspensein 19 of 1179, across 8 files
- Respect prefers-reduced-motion media queriesin 18 of 1179, across 10 files
- Prioritize CSS-only animations for HTMLin 18 of 1179, across 16 files
- Use compound components for related UI elementsin 18 of 1179, across 7 files
Said here and by no other author read
- Pick a workflow from the decision tree
- Execute the checklist for the chosen workflow
- Measure time to first audio
- Show model limits in the user interface
- Verify dataset legal and diversity before scaling
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.