THREAT // 01HIGH
Tanner
A three-channel threat: main rush, motel-room event, and camera jump.
- Click gate
- 350
- Base roll
- 45%
RESEARCH STATEBUILD-BOUND
UNOFFICIAL SURVIVAL REFERENCE / LOCAL EVIDENCE ONLY
Know what
hunts you.
The numbers, timers, and state paths hidden inside Welcome to the Game III— decoded into a field manual you can use before the next knock at the door.
OPEN CASES03EVIDENCE CLASSES03NETWORK SOURCESNONE
CURRENT COVERAGE
Open case files
Every claim is marked by confidence, so decoded values never blur into interpretation.
THREAT // 02SEVERE
Lucas
Patrol-and-investigate AI with several contextual kill states.
- Lock delay
- 10–15s
- Patrol stops
- 6–7
THREAT // 03VARIABLE
Cletus
A distinct key-scaled threat. His numbers must not be attributed to Lucas.
- Base roll
- 35%
- Spawn timer
- 15–25m
RULE // 001
A chance is not a clock.
Percentages below apply to an eligible event check. A 45% Tanner value does not mean a 45% chance every second.
THREAT DOSSIER // 01
Tanner
PRIMARY GATE350player clicks
● Verified asset values◆ Reflected code symbols△ Behavioral inference
INTERACTIVE TABLE
Spawn roll calculator
50%
4-key override
Per eligible main-rush check.
ANOMALY: The raw asset contains 900. The effective meter is capped at 100.
CONTROL FLOW
Three independent channels
Main rush
PlayerClickCount is compared against the configured 350-click gate before the rush path can roll.
Motel room
Own roll, key overrides, and MotelRoomSpawnOnCD state. It does not share the camera cooldown.
Camera jump
A constant 60-style chance plus CamJumpOnCD and its own saved remaining time.
RAW CONFIG
Difficulty comparison
| Path | Normal base | Normal overrides | Leet base | Leet overrides |
|---|---|---|---|---|
| Main rush | 45 | 4:50 · 5:50 · 6:55 · 7:60 · 8:60 | 70 | 4:70 · 5:70 · 6:80 · 7:90 · 8:100 |
| Motel | 42 | 6:50 · 7:55 · 8:55 | 70 | 6:80 · 7:900 · 8:100 |
| Camera | 60 | none | 60 | none |
MITIGATION // VERIFIED
Delay the main check
The only directly exposed main-path gate is click count. Avoiding unnecessary clicks delays eligibility; it does not disable Tanner.
MITIGATION // VERIFIED
Treat channels separately
A motel cooldown is not protection from a camera jump or main rush. Each route owns independent state.
MITIGATION // INFERRED
Respect the safe buffer
The rush actions include parking lot, lobby doors, leaving the safe buffer, panic release, soft despawn, and despawn.
THREAT DOSSIER // 02
Lucas
LOCK DELAY10–15seconds
IMPORTANT DISTINCTION
No ordinary Lucas spawn table exists.
Lucas is connected to simulated-player triggering. The nearby key-scaled 35% table belongs to Cletus.
NORMAL MODE
Behavior timings
Lock picking10–15s
Patrol wait8–12s
Patrol stops6–7
Minimum spots2
Investigation10–15s
DOSCoin stolen25–60%Leet: 65–90%
SPAWN PATH
Simulated-player trigger
AI_CanSimulatedPlayersTriggerLucasAI_TriggeringLucas_CoolDownWindowAI_TriggeringLucas_ChanceToFireAI_TriggeringLucas_Fee
The field names prove the route exists. Its numeric roll remains unverified because this tail is packed through an inherited unversioned schema.
STATE MACHINE
Observed action vocabulary
ROAMpatrol plan
→HEARD NOISEevent
→INVESTIGATE10–15 sec
→PLAYER CAUGHTcontextual kill
Staged, standard, and hard computer-kill actions exist.
Two behind-kill variants are reflected.
Front and back-door kill actions are present.
Dedicated locker and laundry kill actions mean hiding is contextual, not absolute.
MITIGATION // VERIFIED
A lock buys time
Lucas has a 10–15 second lock-picking window. A locked door is a delay, not a permanent barrier.
MITIGATION // SYMBOL
Zannac helps hiding
Lucas_HideGameZannacDifficultyReduction explicitly reduces hiding-game difficulty.
MITIGATION // INFERRED
Break investigation
Noise transitions Lucas into investigation. Relocate during the delay instead of remaining at the noise source.
THREAT DOSSIER // 03
Cletus
NORMAL BASE35chance value
NORMAL
Key scaling
1337 / LEET
Key scaling
Release window12–25m
Spawn window10:00–18:20
SEPARATE ACTOR
Cletus ≠ Lucas
Cletus owns the normal 35% key-scaled table. Lucas owns patrol variables and a simulated-player trigger route.
RESEARCH NOTES // 05
Methodology
EXTERNAL SOURCES0local reverse only
Cooked asset extraction
The active IO Store difficulty assets were converted to legacy Unreal packages without changing the installed game.
Unversioned header decoding
Serialized property fragments were reconstructed to determine which fields override class defaults in Normal and Leet modes.
Reflection metadata
Property parameter tables in the executable supplied field names, types, nested map layouts, and internal offsets.
Behavioral vocabulary
Manager state and action enum symbols reveal possible transitions. They are evidence of available behavior, not proof that every branch is reachable.
● Verified
Read directly from the current difficulty asset with a confirmed serializer and field alignment.
◆ Symbol
A reflected property, manager field, or action name exists in the executable.
△ Inferred
A practical interpretation supported by names and state relationships but not yet confirmed in decompiled control flow.