---
name: ui-demo
description: "Show a visible change working, in a narrated, captioned video of the browser (or a screenshot when one frame shows it), attached to the reply that links its PR. Use whenever a change touches what people see."
title: ui-demo
canonical_url: https://skilld.dev/gh/vercel/board/ui-demo
last_updated: 2026-11-19T06:15:42.000Z
---

> Index of every page on this site: [llms.txt](https://skilld.dev/llms.txt)

> **Skill from skilld.dev.** Follow the user's requested action.
> If the user asked to fork this Skill, follow the fork workflow below. Do not execute the Skill instructions while copying it.
> Otherwise, follow the instructions below for this session. You do not need to install anything.
>
> Supporting files, fetch one when the Skill refers to it: [scripts/ui-demo](https://skilld.dev/api/skills-raw/vercel/board/ui-demo/scripts/ui-demo).
>
> If the user asked to install this Skill, run `npx skilld install vercel/board/ui-demo`. Install writes the Skill files into the project, so every session loads them.
>
> ## Fork workflow
>
> A fork creates an editable local Skill with its original author and licence. The request authorizes copying and local installation.
> 1. Check `./skills/ui-demo`, the project lockfile, and selected Agent targets together. If the local directory or installed Skill exists, stop. Never overwrite an existing directory or Agent target.
> 2. Read [source metadata](https://skilld.dev/api/v1/skills/vercel/board/ui-demo) once. Use sourceUrl, sourceCommit, skillPath, sourceGone, and license. If the source is gone or its path is missing, stop. If license is null, read licence files at the source commit.
> 3. Fetch only the source commit into a temporary Git repository. Do not clone full history. Derive repository_url from sourceUrl, including repository renames. If sourceCommit is absent, resolve the sourceUrl ref once. Set source_commit to that actual commit. Run these commands in one shell call:
>
> ```sh
> git init --quiet "$temporary_dir"
> git -C "$temporary_dir" fetch --quiet --depth=1 "$repository_url" "$source_commit"
> git -C "$temporary_dir" checkout --quiet --detach FETCH_HEAD
> ```
>
> Read applicable licence declarations and notices at that commit. If copying is not permitted, report the restriction and stop.
> 4. Inspect source entries together, then copy the directory containing skillPath into `./skills/ui-demo`. Use the user's path if selected. Keep the original SKILL.md, relative links, scripts, binary assets, and executable modes. Exclude .git metadata. Reject symlinks and paths outside the Skill directory. After checking entries, use cp -a where available. A regular source directory needs no custom copy script. Do not save this page wrapper as SKILL.md.
> Preserve author credit, notices, and applicable licence files from repository or parent directories. Add PROVENANCE.md with the Skill page, source URL, actual commit, original path, and licence. Retain any existing PROVENANCE.md and record new provenance separately. Batch source inspection, copying, and provenance work where practical.
> 5. In the project root, run `skilld install ./skills/ui-demo --mode copy --plain`. If skilld is unavailable, use `npx skilld install ./skills/ui-demo --mode copy --plain`. This known command needs no help lookup. Install does not support --json. Use detected Agent targets, or add --agent for the targets the user selected. Install the local path, never the upstream selector. If installation fails, preserve the local copy and report the exact failure.
> 6. Confirm the local lockfile source and installed Agent copies once. Report the local path, actual commit, and Agent targets. After edits, reinstall the same local path. Upstream updates must not replace it. Do not publish or push unless the user asks.

# Show a change working

When a change touches what people see, show it instead of describing it: a video for a flow
people click through, a screenshot when one still frame shows it, or nothing when there's
nothing to see (say so in the PR instead).

The video records the browser you drive with the `browser__*` tools. Before each action you
call `narrate` with one line, which is said out loud and captioned from that moment, and the
action follows once it's been said. `scripts/ui-demo` in this skill records the tab, cuts the
time you spend thinking, burns the captions in and mixes the voice, as an mp4 under 4 MB.

## 1. Pick where to try it

- **A dev server in the sandbox**, almost always: start it in the background
  (`pnpm dev > /tmp/dev.log 2>&1 &`, after `pnpm install`) and poll it with curl until it
  answers. Read the repo's AGENTS.md or README for how it runs and what it needs. When it
  doesn't run yet, the try-it skill says what you can get, and what you can't.
- **A deployment**, when the change needs real data or the deployed setup and the deployment is
  open to the public. A protected preview asks for Vercel login, which the browser can't do.

## 2. Rehearse, then record

Do the flow once without recording, so the take has no dead ends: `browser__open` the page,
`browser__snapshot`, act on its refs. Then go back to where the flow starts, and:

1. `scripts/ui-demo start`, with the page already open in the browser.
2. For each step: `narrate` its line, then do it with the browser tools. Lines are short,
   first-person sentences people can hear in about 2 seconds, written to be said (words, not
   symbols or paths): "Opening the settings page", "Turning on dark mode", "The page switches to
   dark: it works".
3. `scripts/ui-demo stop /tmp/<thread>-<flow>.mp4`

- Only narrate a result ("it works", "the message is posted") once a snapshot or screenshot shows
  it. If the flow fails, record the failure and say so. Don't attach a video that claims something
  the screen doesn't show.
- The take keeps each line and a few seconds after it, and cuts the rest, so take your time.
- `scripts/ui-demo start --mute` records captions only, when sound adds nothing.
- If `stop` says the video is over 4 MB, record a shorter take.

## 3. Attach it

Attach it to the reply that links the PR: `vb reply <id> "<the PR link, and a line on what the
video shows>" --attach /tmp/<thread>-<flow>.mp4`. A screenshot works the same way
(`browser__screenshot`, then `--attach`). Never attach test logs: the tests are in the PR.
