Night At The Infirmary Anomalies

Night At The Infirmary is a Roblox beta horror experience by A.D. Games where you work a night shift at a school infirmary and decide which visitors are genuine human patients and which are dangerous anomalies in disguise. Anomalies mimic people well enough to pass a casual glance, but the game forces you through a deliberate three-layer verification loop before you ever raise the security shutter. Miss a tell on any layer and the shift can end in seconds.

This hub collects every anomaly reference on the wiki. Unlike a single checklist, these pages group tells by where you spot them: at the reception window, on the printed photo, or through the CCTV monitor bank. Some entities look almost normal in person yet reveal wrong eyes on camera or a cursed grin only on the print. Others fail the window check immediately with hollow sockets or teeth that do not belong on a human face.

If you are new to the game, start with our how to detect anomalies guide for the full inspection workflow, then return here when you need to identify a specific tell. Pair these pages with the interactive anomaly checklist until the three-layer order feels automatic during busy nights.

Every patient in Night At The Infirmary must pass three independent checks before you admit them. The layers are designed so that no single viewpoint tells the whole truth — a patient can look fine at the window, fail the photo comparison, and still look different again on CCTV. Your job is to compare all three and reject on the first confirmed mismatch using the shutter button near reception.

1. Window visual check — Inspect the live patient through the reception glass. Scan eyes, teeth, skin tone, posture, proportions, and movement. Many anomalies expose themselves here with wrong eye count, unnatural stillness, or grins that do not match human anatomy. See our dedicated visual tells guide for the full list.
2. Photo print check — Snap an identification photo and compare the printed still against the patient standing in front of you. Photo-only tells include feature mismatches, grainy static textures, and cursed images with bloodshot eyes. Some prints show horns, extra eyes, or ear shapes that were not visible at the window. Read the photo tells page before your first night.
3. CCTV camera check — Switch to the security feeds covering the entrance, hallway, and waiting area. Camera-only anomalies include void-black bodies, stretched limbs, black boxes over eyes, and figures that stare directly into the lens. A patient who passed the first two layers can still fail here. Use the camera tells guide for feed-specific red flags.

Reject the patient with the shutter if any layer confirms an anomaly. Only admit when window, photo, and CCTV all agree the visitor is human. For decision timing, see how to reject patients.

The wiki splits anomaly knowledge into focused pages so you can study one inspection layer at a time or look up a specific category during a shift break. Each page lists confirmed tells from beta community reports under UPD 1, with notes on which layers they appear on and what action to take.

• All Anomaly Types — Master index of every documented anomaly category: visual-only, photo-only, camera-only, and multi-layer tells. Use this when you are not sure which page to open.
• Visual Tells — Window-layer red flags including hollow eyes, sharp teeth, wrong proportions, twitching, and head tracking.
• Photo Tells — Print-layer mismatches, static photos, cursed grins, and feature differences between the live patient and the desk snapshot.
• Camera Tells — CCTV-only distortions, void bodies, skinwalker shapes, camera staring, and sanity-draining feed events.

For step-by-step application during gameplay, combine these references with the first night walkthrough and the how to beat anomalies guide when nights escalate.

The golden rule across every anomaly page is simple: one confirmed tell on any layer means reject. You do not need two or three matching signs — a single clear mismatch between the window view and the photo, or a void-black silhouette on CCTV alone, is enough to hit the shutter. Hesitating because a patient "mostly looks human" is how beta players lose shifts.

Admission is the absence of rejection. When all three layers align — eyes match between window and photo, proportions look human on CCTV, no static or cursed print artifacts, no camera-only distortions — let the patient proceed. Wrong rejects carry consequences in some nights, so the three-layer system exists to give you confidence before you decide, not to encourage guessing.

Special caution: some photo and CCTV events drain sanity when you stare at them. Cursed photos and feed zooms toward dark figures punish prolonged looking. That is why experienced players check the window and CCTV before taking a photo when they already suspect an anomaly — there is no reason to trigger a sanity hit on a patient you plan to reject anyway. Learn the full control flow on our controls page.

Night At The Infirmary remains in beta under UPD 1. A.D. Games may add new anomaly types, rebalance tell frequency, or introduce additional inspection layers in future patches. We label community-verified tells separately from single-report rumors on each subpage and refresh content when UPD 1 patch notes confirm changes.

Anomaly difficulty scales across nights. Early shifts teach basic visual tells; later nights stack photo mismatches and camera-only skinwalkers that look normal at the desk. The all nights walkthrough describes when new tell families tend to appear so you can pre-read the right reference page before starting a shift.

Community findings also live on our Discord and Trello pages for raw beta reports. The anomaly hub here is the curated, gameplay-focused version — start here, drill into subpages, and use guides when you need procedural help rather than tell identification alone.