AddThisFeature

Visual Asset Optimization Pipeline

Turn source artwork into small, correct derivatives without losing the detail that matters.

involved Front-End Performance

What it adds

A repeatable processing step that derives sized and compressed variants for every image, icon, and illustration the app ships.

What your agent is told to do

5
  1. 1

    Locate every visual asset the app serves — bundled artwork, icons, illustrations, and user uploads — and bring them all through one processing path instead of optimising some by hand.

  2. 2

    Keep the untouched original of every asset in a place the pipeline can reach, so a change in format or quality target can be applied by reprocessing rather than by re-sourcing artwork nobody can find.

  3. 3

    Set the compression target per kind of asset. A photograph tolerates aggressive compression; a screenshot containing text and a flat illustration with hard edges show artefacts at the same setting.

  4. 4

    Emit a manifest describing every derivative produced — its dimensions, format, and the original it came from — and let the Adaptive Media Source Component read that manifest rather than construct URLs for files the pipeline may never have made.

  5. 5

    Do not reprocess everything on every build. Key derivatives by the content of the source so an unchanged asset is reused, or the pipeline becomes slow enough that someone disables it.

Edge cases it handles

7
  • Originals must survive. Once the pipeline overwrites its own input, the next change of format or quality has nothing to work from and the artwork has to be recovered from a designer's machine.
  • One quality setting cannot serve every kind of asset. Verify a photograph, a text-bearing screenshot, and a flat vector-style illustration side by side before fixing the defaults.
  • Transparency and animation must survive conversion. An alpha channel flattened to white and an animated file reduced to its first frame are both silent failures that only show up on the page.
  • Every derivative must be checked before it is published — that it opens, that its dimensions are right, and that it is genuinely smaller than the source. A conversion that produced a larger file should be discarded in favour of the original.
  • Orientation metadata must be applied and then stripped, so a photograph does not publish sideways, and location and camera metadata never reach a public URL.
  • Colour profiles must be converted rather than dropped, or brand colours will shift between the source file and the rendered page.
  • User uploads have to travel the same path at runtime, with a size and dimension limit and a defined behaviour for a file the pipeline cannot process.

Definition of done

9
  • Every visual asset the app serves is produced by the pipeline rather than optimised by hand.
  • Untouched originals are retained and reprocessing regenerates all derivatives from them.
  • Compression targets differ by asset type and were validated against a photograph, a screenshot, and an illustration.
  • Transparency and animation are preserved in every derivative that claims to support them.
  • Each derivative is verified for openability, dimensions, and size before publication, and a larger-than-source result is rejected.
  • Published files carry no camera or location metadata.
  • An unchanged asset is not reprocessed on a rebuild.
  • 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.