AddThisFeature

Video Thumbnail Picker

Choose the frame that represents a video, or upload a cover image instead.

involved Media & Video

What it adds

A picker that offers candidate frames from a video and accepts a custom uploaded cover image.

What your agent is told to do

5
  1. 1

    Generate candidate frames on the server as part of the video's processing pipeline, and show the user a strip of stills to choose from. Do not stream the whole file into the browser to seek it frame by frame.

  2. 2

    Offer a scrubber for picking an exact moment in addition to the suggested stills, so someone who wants a specific frame is not limited to the ones you guessed.

  3. 3

    Accept a custom cover image upload as an equal path, and route it through the app's existing File Upload and Image Compression handling rather than adding a second upload endpoint with its own limits.

  4. 4

    Store one canonical still and derive the sizes each surface needs: grid tile, card, and social share preview. Reuse the app's existing image derivative and cropping behaviour, including Image Focal Point Cropping if it exists, so a face is not cropped out of the square version.

  5. 5

    Do not block publishing on thumbnail selection. Pick a reasonable default automatically and let the user change it later, or half your videos will sit unpublished behind an unmade decision.

Edge cases it handles

8
  • Videos often open on black, a fade-in, or a title card. Score candidate frames and skip near-blank or single-colour ones when suggesting defaults, or every automatic thumbnail will be a black rectangle.
  • Replacing the source video invalidates every stored still. Regenerate candidates on replacement and clear the previous selection rather than leaving a cover image from a video that no longer exists.
  • Different surfaces need different aspect ratios, and a still cropped independently per surface drifts. Store consistent, named sizes generated from one source frame so the grid, card, and share preview agree.
  • A thumbnail can be requested while the video is still encoding. Show a pending state on the picker and generate candidates when processing completes, instead of failing or serving an empty image.
  • An uploaded cover image is user-generated content on a public page. Enforce type and size limits on the server, run it through whatever scanning the app already applies to uploads, and treat it as moderatable.
  • Reserve the thumbnail's dimensions in every layout that shows it so cards do not resize when images arrive.
  • Every thumbnail needs alternative text. Default it to the video title rather than leaving it empty or filling it with a filename.
  • Generating stills for a long or high-resolution video is expensive. Cap the work per video and let it fail gracefully to a single default frame rather than tying up the encoder.

Definition of done

9
  • Candidate frames are produced server-side and offered as a selectable strip.
  • A user can also seek to an exact frame or upload a custom cover image.
  • Near-blank frames are excluded from automatic suggestions.
  • Replacing the source video regenerates candidates and clears the stale selection.
  • One selected still produces consistent derivatives for grid, card, and share surfaces.
  • A video still encoding shows a pending picker rather than an error.
  • Uploaded covers pass the app's existing size, type, and scanning rules and carry alt text.
  • The feature matches the existing design system.
  • No existing functionality is broken.

Related features

How it works

  1. 1

    Copy the link

    Grab the Markdown instruction URL for this feature.

  2. 2

    Give it to your AI

    Paste it into Claude Code, Cursor, v0, Lovable — whatever you build with.

  3. 3

    It inspects, then implements

    Your agent reads your existing app first, then adds the feature to fit it.

Works with your stack

These instructions are written to adapt. They tell the agent to detect your framework, match your existing design system, and reuse what you already have — rather than assuming a particular stack.

Need it tighter than that? Customize the feature and tell it exactly what you're running.