Maintaining the cafe is part of Anomaly Cafe’s official premise, but the creator does not publish a fixed task list, trigger schedule, cooldown, or penalty table. Treat maintenance as a visible operational problem: read the current prompt or state change, decide what it blocks, assign one owner, and confirm that the station or route works again. This keeps the guide useful when alpha details move.
Triage by player impact
Handle a problem immediately when it blocks a required preparation station, prevents delivery, narrows the team’s safe movement, hides an observation reference, or appears in the active objective. A cosmetic-looking issue with no current impact can wait until a natural transition. Do not accept more tickets when the tool needed to complete them is unavailable.
Use four questions: What changed? Where is it? Which current task cannot proceed? What does the live game ask you to do? This is safer than assuming every spill, prop, sound, or animation has the same meaning in every build.
Assign upkeep without losing service
The runner or watch/floater is usually the best maintenance owner because they already move between zones. Intake pauses only if the issue affects the queue or safe response. Preparation finishes a safe current step and avoids starting work that depends on the blocked area.
Announce the problem and ownership: “preparation route blocked, runner handling.” When it is fixed, announce “route clear.” Those two messages prevent duplicate repair attempts and tell service players when they can resume normal flow.
Maintenance priority board
Separate upkeep from anomaly evidence
A moved object, changed light, unusual sound, or broken route may be maintenance, an anomaly cue, or an alpha bug. Do not erase evidence before communicating it. State the location and what changed, compare with the current objective or warning feedback, and ask one teammate to confirm. If the game presents an upkeep interaction, handle it. If danger feedback appears, switch to the anomaly response route.
This separation matters because a tidy baseline makes real changes easier to notice, but overconfident cleaning can also remove the team’s chance to compare what happened.
Recover after an interruption
When danger ends, walk the critical loop: intake, preparation, delivery, observation, and regroup routes. Check current prompts and visible blockers. Clear ambiguous items and assign unfinished maintenance before reopening intake. Start one ticket and confirm that all required stations work.
If a station remains broken, reduce the service plan rather than stacking impossible requests. Re-read the objective, ask whether another player owns an interaction, and rejoin only after preserving the context needed for a useful bug report.
Solo maintenance routine
Keep one active request. After delivery, scan the route and current prompts. Clear a blocker before taking the next task. During danger, abandon nonessential upkeep and preserve a safe path. Afterward, inspect the station needed by your next request first.
Solo players cannot cover every area at once, so use predictable check points instead of constant wandering. A short loop after delivery is more reliable than reacting to every distant sound.
Team maintenance routine
Assign a mobile player and define their territory. Intake reports queue impact; preparation reports station impact; the runner reports path impact; the watcher reports baseline impact. The maintenance owner chooses the current highest priority and closes it with a clear message.
Rotate ownership during a calm period so one disconnect does not remove the only player who understands upkeep. In a large lobby, split by zone, not by vague task names that may change.
What remains update-sensitive
The complete list of upkeep interactions, their triggers, exact controls, cooldowns, failure consequences, and relationship to anomaly events require current live-game confirmation. Community guides checked on August 20, 2026 describe clearing dishes, spills, furniture, or other blockers as part of keeping the shift stable, but those examples should not be treated as an exhaustive official list.
Use the visible prompt and result. If a reported task does not exist in your build, do not force it or label the server broken.
Troubleshooting
If an interaction does nothing, confirm that the objective targets that problem, that you are in range, and that another player is not holding the required state. If the task returns immediately, look for a second linked blocker and compare with current warning feedback. If only one server reproduces the issue, it may be stale or bugged; record the context before moving.
Continue with Recipes and Serving to balance upkeep against the queue, Map Guide to protect functional routes, and How to Spot Anomalies when a change may not be ordinary maintenance.