Design reviews argue about screenshots. Meanwhile the thing your users actually load has quietly moved. A radius that isn't quite the token, a focus state that never shipped, an empty screen nobody has looked at since the release it went out in.
Your system is probably in good shape wherever it's defined. It's the built product that drifts, and checking all of it is tedious enough that it doesn't happen. Strawberry works in the browser, so it can go and look at the real thing: the pages that render, in the states that actually break, across the breakpoints and themes you care about.
What Strawberry can help you do
Not sure where to start? Tell your companion what you're working on and it can suggest a first step. See the skill
Find where it has drifted
Drift is invisible one page at a time and obvious in aggregate. A single button with the wrong radius is nothing. The same button rendered four different ways across one signup flow is a system problem, and you only see it if someone walks the whole thing.
That walk is tedious, which is why it doesn't happen. Your companion can do it against the live product, in the states that actually break:
- Spacing that isn't on the scale, and type outside the ramp.
- Colors that are nearly a token but not the token, and text that fails contrast on its real background.
- The same component built three ways, or a variant used somewhere it wasn't meant for.
- Focus that isn't visible, hover that does nothing, disabled that doesn't look disabled.
- Empty and error states, where drift concentrates because they get built last and reviewed least.
Findings come back grouped by cause rather than by page. Twelve instances of one wrong token is a single fix, and it should read like one. Accessibility failures come first, then anything on a primary flow, then the cosmetic drift.
It also won't file a deliberate exception as a bug. Where something looks intentional, it asks rather than assuming.
Audit the system
Walk agreed pages and record where the built interface has left the system, with evidence you can inspect.
Audit the flow
A different question: not whether it is consistent, but whether it actually works.
Ground it in what users said
Design reviews go badly when the strongest evidence in the room is a preference. Themes pulled from actual feedback, with the quotes still attached, tend to end the argument faster than another round of opinions.
Turn feedback into themes
Source-linked patterns and tensions from feedback you already have.
Understand the audience
A sourced picture of who you are actually designing for.
Hand it over cleanly
A design decision that lives only in a file gets rebuilt from memory. Turning it into a spec, or into something presentable for the people who have to agree to it, is the part that decides whether it survives contact with a sprint.
Specify it
Turn the accepted decision into something the people building it can follow.
Present it
Move from a file to something a room can follow and respond to.
Make reviews stick
Design decisions get made in a room and then quietly evaporate. Three weeks later the same debate runs again, because nobody wrote down what was settled or why.
Walk in prepared
What was decided last time, what has changed since, and the questions you need answered.
Leave with it written down
The decisions, the open questions, and who owns the next move.
Keep the product honest
Drift comes back. It always does, because it's the by-product of shipping quickly rather than of anyone being careless.
Once an audit is right, the system it checks against, the surfaces it walks, and the states it covers can be saved. If the whole team should be checking the same way, that method can be shared rather than living with whoever set it up.
After that, a Routine can re-walk the agreed surfaces once a release has landed and tell you only what is newly broken.