Runtime debugging and decision history
Select the correct running Actor, inspect execution, and pin an earlier frame when you need to explain a choice.
Attach to a running AI
During Play mode choose the Actor in Running AI. Open its root graph and double-click through BDS, Ability and behavior references. Back / Forward retraces the visited path. With multiple Actors, make sure the selected instance matches the graph you are inspecting.
Breakpoints and stepping
| Control | Purpose |
|---|---|
| F9 / node debug menu | Toggle a breakpoint on the selected node. |
| F5 | Continue or pause debugging. |
| F10 | Step a node execution boundary; do not force a persistent Action to finish. |
| F11 | Advance one decision/execution frame. |
| Green breathing outline | The node actually executing. |
| Red debug outline | Breakpoint or paused debug location, distinct from a failed condition. |
Read-only mode permits navigation, selection and breakpoints. Unlock editing to modify the graph, then apply to the selected runtime instance. Saving configuration resources remains a separate action.
Review the last 100–300 frames
Choose capture target and capacity
Select an AI, enable recording and choose 100, 200 or 300 frames.
Scrub time
Disable Follow latest and drag to an earlier frame. Read that frame’s values, leading candidate and executing candidate.
Inspect a child BDS
Double-click a child BDS execution or select it in the hierarchy. Navigate Back / Forward along the inspection path.
Return to live observation
Enable Follow latest. Browsing history does not rewind the game world.
Keep capture bounded
Record the relevant Actor only and disable history when finished. Compilation, leaving Play mode or clearing history may discard records. The bounded ring is not a permanent replay file.