chore/atlas-100

needs attention

dde5c8c · fullpre-PRreviewed 2026-08-03 19:19 UTC0H · 1M · 0L · 2I

The branch

Purpose

Increase PostHog session replay coverage for Atlas to 100% of consenting sessions for a time-boxed experiment

Goal

Capture every consenting Atlas session during the experiment window to maximise replay data quality

Sub-goals

SG-1: Bump sampleRate from 0.6 to 1 while keeping requireConsent: true
SG-2: Document the temporary nature and revert path in the comment

The changes (whole branch)

What: PostHogProvider sessionReplay.sampleRate changed from 0.6 to 1 in apps/web root layout; comment updated to explain time-boxed nature
Why: Time-boxed experiment: previous 60% rate was a steady-state default, not tuned for experiment coverage
Areas: apps/web/src/app/layout.tsx+9−6
Blast: 1 file, +9/-6 lines; marketing website root layout only

time-boxed experiment revert required after experiment ends (BAT-292 follow-up)

➖ci· No GitHub credentials available in this environment➖coderabbit· No .coderabbit.yaml in repo

Findings · 3

security1

info

Consent gate remains intact at 100% sampling

apps/web/src/app/layout.tsx

requireConsent: true unchanged — no recording without explicit opt-in.

conventions1

info

9-line comment is verbose but justified

apps/web/src/app/layout.tsx:71

Captures 5 distinct non-obvious constraints; satisfies CLAUDE.md 'explain WHY' rule.

improvement1

medium

Externalise sampleRate to NEXT_PUBLIC_POSTHOG_SESSION_REPLAY_SAMPLE_RATE env var

apps/web/src/app/layout.tsx

This change needed a commit; revert will need another. An env var (defaulting to 0.6) lets future experiments be tuned via Vercel env vars without touching code.