Questions, answered honestly.

Into a sandboxed app folder on your iPhone, not your Camera Roll (unless you explicitly export them), and never to our servers. Face detection during capture and import uses Apple's Vision framework on-device. Image bytes never leave the device.

Severity scores, trigger tags, timestamps, and app preferences sync through Apple's CloudKit using your Apple ID, in your private container. This is automatic when you're signed into iCloud, and it's how your timeline survives a phone swap. We can't read your CloudKit container; only your Apple ID can. Your photos are not part of this sync.

Two operations-only SDKs ship in the binary: TelemetryDeck for aggregated app-usage signals (no per-user storage, no session replays, hashed user IDs) and Sentry for crash reports (screenshot and view-hierarchy attachments off). Both are subject to a strict allowlist enforced at the call site: lifecycle, auth, and paywall events with non-content properties only. We never send photo metadata, severity scores, trigger names, Fitzpatrick selection, or any free-text. No Facebook pixel, no Google Analytics, no advertising SDK. The privacy page documents this in full.

Two paths. Deleting the app removes everything on the device. The "Delete account" row in Settings wipes your CloudKit private record zone first; only after that confirms does it wipe local data and sign you out. CSV self-export is available at any time, regardless of subscription state, in case you want to keep your history.

Skinframe needs at least three flare days that share a recent trigger before it surfaces a pattern. With consistent logging that's usually two to four weeks. Until then, Insights shows what it's still watching: the largest sub-floor count for each candidate trigger, and how many more co-occurrences are needed to surface it. No pattern is shown without the count behind it.

You score it. The daily form mirrors the PSA per-feature scale: 0 to 4 for erythema and burning/stinging, 0 to 3 for papules and telangiectasia, plus a transient flushing toggle and ocular symptom multi-select. There is no global composite score by design, the literature shows composite scoring obscures phenotype differences. Skinframe does not estimate severity from the photo; the photo is for your visual record and for the dermatologist handoff.

Per-trigger lag windows tuned to the physiology: alcohol and hot drinks 12 hours, sun and heat 12 hours, stress and sleep and cycle 72 hours. Pattern detection counts distinct flare days that share a trigger within the window. Results require at least three matching flare days before they surface, which is the floor that keeps coincidence from reading as signal. Every insight shows the count, the window, and a "co-occurrence is not causation" footnote. We don't hide the math, and we don't run hypothesis tests we can't interpret.

One US Letter page: a slate header band with the period range, a period summary (flare-day count and average severity in serif numerals), the top three lag-aware co-occurrences with their counts and windows, calmest and worst photos for the period, an optional RosaQoL block if you've completed it, and a methodology footer that names the consensus framework (ROSCO 2017 + NRS 2017) used for severity scoring. Generation timestamp tagged.

Yes. The dermatologist PDF is a Skinframe Pro feature. CSV self-export of all your data is available without a Pro purchase, regardless of subscription state, so you always have a way to take your history with you.

Skinframe is free to install and use for fourteen days. After the trial, continued use of capture, import, lag-aware correlation, and the dermatologist PDF requires Skinframe Pro: $29.99 one-time lifetime, or $4.99/month auto-renewing. Identical features either way. Read-only access to your existing timeline, photos, and severity history remains available after the trial without any purchase, and CSV self-export is always available so you can take your data with you. One-tap cancel via Apple, no auto-renew rate hikes, no email nag.

The lifetime is the recommended path. It pays back in six months versus the subscription, and once you own it the price never changes. The subscription exists for people who want to commit a smaller amount up front. There are no feature differences and no premium-tier carve-outs.

Skinframe leans on Apple's Vision framework for face detection and on private CloudKit containers for sync. We'd rather do iOS well than Android adequately, and we won't ship a worse version of the privacy story.

No. Skinframe is a self-tracking and journaling tool. It's not cleared by the FDA, not a substitute for clinical evaluation, and not designed to diagnose, treat, cure, or prevent any condition. Always work with a licensed dermatologist on treatment decisions.

The trigger taxonomy, severity rubric (PSA-derived per-feature scale), and lag windows are tuned to rosacea. People with adjacent inflammatory conditions sometimes find the photo log + co-occurrence engine useful, but the literature substrate behind the patterns is rosacea-specific, so we don't claim accuracy outside that.

Yes, and the literature is clear about it. Rosacea on Fitzpatrick I–III often presents as visible redness; on IV–VI it's more reliably read through sensation: burning, stinging, transient flushing episodes plus visible bumps. Skinframe asks during onboarding so the app can foreground the surface that matches your phenotype: visible-redness severity for I–III, sensory diary first for IV–VI with photo trajectory as the secondary read. References: Maliyar 2022 (PMC9165629), Daneshjou 2022 (Sci Adv DDI dataset), Alexis 2019 (JAAD).