Network requests

Meticulous counts the network requests each session issues as it replays against the code under test, and compares that count against the same session replayed on the base commit. A change that makes your app chattier — an accidental N+1, a dropped cache, a component refetching on every render — often produces no visual difference at all: the page looks identical while quietly issuing far more traffic. This check surfaces those regressions on the pull request, with a per-endpoint breakdown of exactly which requests grew.

Every fetch and XHR request the app issues during replay counts, except traffic to known telemetry and third-party hosts — analytics, error tracking, fonts (Segment, PostHog, Sentry, Datadog, Google Analytics, LaunchDarkly, Google Fonts, and similar). Those are excluded so that an extra analytics beacon can never flag your pull request; requests to your own backend always count. The list of ignored hosts can be customised per project.

By default, a session that issues at least 20% more requests than on base marks the Meticulous network requests status check as failed on the pull request, requiring the author to acknowledge the result in Meticulous before the pull request can proceed. A session with at least 10% more requests is reported as a warning that leaves the check passing. Both thresholds are configurable per project.

The check can also be made informational only: every session with more network requests is then reported as a warning and the status check never fails, so you can see what it finds without blocking any pull request.