Font Loading Strategy
Load the custom fonts without invisible text, broken metrics, or a late reflow.
What it adds
A defined loading, fallback, and failure path for every custom font the interface depends on.
What your agent is told to do
5
What your agent is told to do
5-
1
List every font family, weight, and style the app actually renders, then check that list against what is being downloaded. Most apps ship weights nothing on the page uses.
-
2
Choose a fallback whose metrics are adjusted to match the custom face, so the text laid out before the font arrives occupies close to the same space and the swap does not reflow the page.
-
3
Preload only the faces needed for the first meaningful render. Preloading everything competes with the request for the content itself and delays the thing the user came for.
-
4
Define what happens when the font never arrives: blocked network, a failed request, an unsupported format. The page must remain fully readable in the fallback indefinitely rather than showing invisible text.
-
5
Do not solve a slow font by hiding text until it loads. Blank text on a working page is a worse failure than a brief change of typeface, and it is the failure users report as the site being broken.
Edge cases it handles
7
Edge cases it handles
7- The fallback must be metric-compatible or metric-adjusted, otherwise the swap shifts every line of text and the layout jumps after the page appeared finished.
- Preloading more than the critical faces steals bandwidth from the first render, so the preload list needs to be short and justified rather than generated from the whole family.
- Variable fonts change the calculation — one file may replace six, but only if the app actually uses the axes it carries, and language subsets need their own decision about which are fetched eagerly.
- A complete font failure must leave the page readable in the fallback rather than blank, which means the invisible-text period has to be bounded and short.
- Fonts loaded conditionally for a script the user's content requires must not arrive so late that the text has already been measured and laid out with a wrong-width fallback.
- Icon fonts fail differently from text fonts: an unloaded icon face renders as boxes or as unrelated characters, which is why icons belong in the Icon System rather than in a font.
- Reserving space for text that has not rendered is shared ground with Layout Shift Guard; this feature owns the font metrics and swap behaviour, and that feature owns detecting the movement when it happens anyway.
Definition of done
9
Definition of done
9- Text is readable throughout loading and never invisible for a prolonged period.
- Fallback metrics are adjusted so the swap produces no visible reflow.
- Only the faces required for first render are preloaded.
- Every weight and style downloaded is one the app actually renders.
- A total font failure leaves the interface fully usable in the fallback.
- Variable font axes and language subsets are loaded deliberately rather than wholesale.
- No icon or symbol rendering depends on a custom font loading successfully.
- The feature matches the existing design system.
- No existing functionality is broken.
Related features
Semantic Color Tokens
Semantic Color Tokens
Name colours by the job they do, so changing a theme is one edit rather than a sweep.
What it does
A vocabulary of role-named colour tokens — surface, text, border, accent, danger, success — that components reference instead of literal values.
How it works
- 1 Inventory every literal colour value in the codebase and group them by the job they are doing rather than by the shade they happen to be. Two different greys used as card backgrounds are one role; the same grey used as a background and as a border is two.
- 2 Define the role names once and map each role to a value per theme. A role may resolve to a light grey in one theme and a near-black in another; what must not change is which components reference it.
- 3 Stop shared components accepting raw colour values. A primitive that takes a hex string as a prop guarantees that some product screen will hardcode a colour the themes never learn about.
Copy the prompt
No account needed
Add this feature to my app:
https://addthisfeature.com/x/semantic-color-tokens
Motion Token System
Motion Token System
Give the app one shared vocabulary of durations and easing curves instead of ad hoc numbers.
What it does
A named set of durations and easing curves, each tied to a category of motion, replacing the timing values scattered through the app's components.
How it works
- 1 Inventory every duration and easing value currently written into the app and group them by what they do: something entering, something leaving, something being emphasised, and something changing size or position.
- 2 Name a small set of durations — roughly instant, quick, standard, and deliberate — and a small set of curves, then map each motion category to a pairing. Four or five of each is enough for an entire product.
- 3 Give entrances and exits different treatment. Something arriving should decelerate into place; something leaving should accelerate away and take less time, because nobody wants to wait for a dismissal.
Copy the prompt
No account needed
Add this feature to my app:
https://addthisfeature.com/x/motion-token-system
Responsive Grid Tokens
Responsive Grid Tokens
Express columns, gaps, and spans as shared tokens rather than per-component guesses.
What it does
A token set describing column counts, gutters, and span behaviour at each breakpoint, used by every grid in the app.
How it works
- 1 Find the grids already in the app — card lists, dashboards, form layouts, media galleries — and record how many columns each uses at each size. The reconciled set is normally three or four column counts, not one per screen.
- 2 Define gutters from the spacing scale rather than inventing grid-specific values, so the gap between cards matches the rhythm of everything around them.
- 3 Express spans in tokens tied to the column count, so a component asking for half a row keeps working when the row drops from four columns to two instead of overflowing.
Copy the prompt
No account needed
Add this feature to my app:
https://addthisfeature.com/x/responsive-grid-tokens
How it works
-
1
Copy the link
Grab the Markdown instruction URL for this feature.
-
2
Give it to your AI
Paste it into Claude Code, Cursor, v0, Lovable — whatever you build with.
-
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.