> Meticulous records real user sessions from a web app and replays them against each commit to catch visual regressions.
>
> Full AI-readable docs index (project setup guide + every page as markdown): https://app.meticulous.ai/llms.txt

# AI slop patterns

Meticulous can flag the visual tells of AI-generated UI that a pull request introduces, such as gradient text, purple-to-blue palettes, glowing dark cards, side-tab accent borders, and nested cards.
When a pull request adds one of these patterns to a screen your users reach, the author is notified with a comment and a CI check.
Meticulous reports only **new** patterns, ignoring pre-existing ones.

## How does detection work?

Every time Meticulous takes a screenshot of your application, it scans the rendered web page with [impeccable](https://www.npmjs.com/package/impeccable)'s open-source anti-pattern detector.
It then compares the patterns found on the pull request with those found anywhere in the base run, and reports the freshly introduced ones along with the user flows and screenshots where they appear.
Meticulous scans the screens and states reached by your recorded sessions on every run, with no extra tests to write or maintain.

## Which patterns are tested?

impeccable's rules fall into two groups:

- **AI slop tells**: patterns typical of AI-generated designs, such as gradient text, AI color palettes, dark glows, side-tab borders, nested cards, and overused fonts. New tells with a warning severity require acknowledgement; the rest are reported as warnings.
- **Design quality**: general issues such as low contrast, tiny body text, skipped headings, and cramped padding. These are off by default, since several overlap with the [accessibility check](/docs/built-in-checks/accessibility), and they never require acknowledgement when turned on.

On [Settings → Checks](https://app.meticulous.ai/settings?page=checks), you can turn individual rules off, turn the design-quality rules on, and exclude pages by URL path.
A newly enabled rule starts reporting new patterns once a base run has scanned that rule too.

The detector is impeccable by Paul Bakaus, licensed under Apache-2.0.

## Does the check block merging?

The AI slop patterns check is blocking by default, but it can be configured to never block pull requests.
You can change this yourself on [Settings → Checks](https://app.meticulous.ai/settings?page=checks), no Meticulous admin needed.

## How can AI slop patterns testing be enabled?

Reach out to the Meticulous team at [support@meticulous.ai](mailto:support@meticulous.ai) and we'll turn it on for your project.
The check can optionally be enabled for a subset of users rather than the entire team.
