Plan and test game mechanics

A mechanic needs an observable rule: an action, a change in state and feedback to the player. Before adding more systems, test whether the first action, win condition and restart work together. The existing guides below provide the starting workflow and debugging checks.

Open Mechanics collection ↗

Prepare a game for publishing

Publishing starts with a game that can be tested, understood and maintained. Check the available creator controls, audience settings and safety requirements before a release. Use the publishing guide for the sequence, and the troubleshooting guide for an actual blocked step.

Open Publish collection ↗

BC-0240 · TROUBLESHOOTING

Editorial review 2026-09-27

When a Prompt Problem Is Probably Not Your Prompt

If the failure happens before your instruction is submitted or evaluated, rewriting its gameplay details is unlikely to address that observed boundary. Record what stage is reachable and what evidence shows the prompt was received. This can rule out an unhelpful wording experiment without proving an outage or a particular platform cause.

Symptom

Repeated prompt rewrites are proposed even though the observed failure occurs before wording can affect the result.

Immediate action

Identify the earliest reachable stage and the evidence that the instruction was actually received.

Diagnostic branches

The failure precedes access to or submission of the input.: Use the appropriate access or sending branch instead of changing gameplay prose.
Acknowledged generation stops unfinished.: Use documented recovery rather than inferring a wording defect.
A playable output exists with unclear or incorrect behavior.: Return to observed design and instruction repair rather than asserting a platform cause.

Least-destructive change

Avoid irrelevant paid rewriting experiments while preserving the observed failure boundary.

Retest

The report distinguishes a stage exclusion from a proven root cause.
No receipt or completed generation is inferred without evidence.
The chosen next check can actually address the reachable boundary.

Escalation

Give official assistance the actual boundary and notice when unresolved. This guide does not establish an outage, an account defect or a hidden processing stage.

Working notes

Use existing observations to locate the earliest failure: opening the creation surface, reaching the input, acknowledging the send, generating output or running the result. A screen that fails before input is available cannot tell you whether a more specific game description would have worked.

Separate logical relevance from a diagnosis. The observation may show that prompt wording has not yet become part of the failing step, but it does not distinguish connection, account, device or service causes. Keep those unknown rather than turning probably not the prompt into definitely a server bug.

Do not send alternate paid requests as a control experiment. Compare the stage and exact notice already present, and follow the appropriate access, sending or stopped-generation guidance. If a playable result exists, wording and design clarification may again be relevant; the earlier-stage reasoning no longer applies automatically.

Prepare a concise boundary statement for assistance: what can be reached, what action was attempted and what acknowledgement is absent or present. Include only relevant nonsensitive context. A support packet should not contain a speculative backend diagnosis merely because local wording edits would not affect the observed stage.

Examples

Boundary exclusion: the creation surface cannot be opened, so no game text has been entered. Changing the wording of an unsent prompt does not test that access failure; its actual cause remains unknown.
Different boundary: the request is acknowledged and a game opens, but the goal is ambiguous. Prompt clarity may be relevant here; do not reuse a platform-failure conclusion from an unrelated access incident.