agentleFS
ukanwatukanwat/aaabench/.claude/skills/unreal-behavior-trees/SKILL.md

unreal-behavior-trees

Skill399 starsChanged 3 months agoMITView on GitHub

Build NPC AI in Unreal Engine 5 with Behavior Trees and Blackboards: composites (Selector/Sequence), tasks, decorators, services, and running the tree from an AIController. Use when creating enemy/NPC AI, BT_/BB_ assets, custom BTTask or BTService nodes, or when the user mentions Behavior Tree, Blackboard, AIController, BTTask, decorator, or service.

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use when building enemy/NPC AI: creating a BT_/BB_ asset pair, structuring Selector/Sequence branches, adding decorators (conditions) and …

    line 24
  2. Very specific:

    Use when the project has Behavior Tree (BT_) and Blackboard (BB_) assets and an AAIController.

    line 27

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Unreal Behavior Trees

Author NPC decision-making in UE5 with Behavior Trees driven by a Blackboard: structure the tree with composites, gate branches with decorators, keep state current with services, and run it from an AIController. Targets UE 5.4+.

When to use

  • Use when building enemy/NPC AI: creating a BT_/BB_ asset pair, structuring Selector/Sequence branches, adding decorators (conditions) and services (periodic updates), writing custom BTTask/BTService nodes, or wiring an AIController to run the tree.
  • Use when the project has Behavior Tree (BT_) and Blackboard (BB_) assets and an AAIController.

When not to use: the concept of AI (FSM vs BT vs steering, cross-engine) → game-ai. Pure navigation/pathing math is engine navmesh (BT's MoveTo uses it). Simple one-off logic may be cheaper as a small state machine than a full tree.

Core workflow

  1. Create the pair: a Blackboard (BB_) holds typed keys (the AI's memory: TargetActor, LastKnownLocation, bIsInvestigating); a Behavior Tree (BT_) references that Blackboard.
  2. Possess and run. An AAIController possesses the pawn and calls RunBehaviorTree(BT), which also initializes the referenced Blackboard.
  3. Structure with composites. Selector runs children left→right until one succeeds (priority/fallback: "attack, else chase, else patrol"). Sequence runs children until one fails (do-all: "move to cover → reload → peek"). Simple Parallel runs one main task alongside a secondary.
  4. Gate branches with Decorators that read Blackboard keys (e.g. "Has Target?" guards the combat branch). Set Observer Aborts so the tree re-evaluates when the key changes.
  5. Keep the Blackboard current with Services attached to a branch — they tick periodically (e.g. update TargetActor via a sight check) only while that branch is active.
  6. Do work in Tasks, which return Succeeded, Failed, or InProgress (latent tasks like MoveTo finish later).
  7. Verify with the Behavior Tree debugger during PIE — it highlights the running node and shows live Blackboard values, so you see exactly which branch executes.

Patterns

1. AIController that runs the tree (C++)

void AEnemyAIController::OnPossess(APawn* InPawn)
{
    Super::OnPossess(InPawn);
    if (BehaviorTree)                 // UPROPERTY(EditAnywhere) TObjectPtr<UBehaviorTree>
        RunBehaviorTree(BehaviorTree); // initializes & uses the Blackboard the BT references
}

2. A priority tree (node structure)

ROOT
└── Selector (try combat, else investigate, else patrol)
    ├── Sequence            [Decorator: Blackboard 'TargetActor' Is Set, Observer Aborts: Both]
    │     ├── Task: MoveTo (TargetActor)          // latent: returns InProgress then Succeeded
    │     └── Task: Attack
    ├── Sequence            [Decorator: 'LastKnownLocation' Is Set]
    │     ├── Task: MoveTo (LastKnownLocation)
    │     └── Task: Wait (3s) + clear key
    └── Task: Patrol (BTTask_FindPatrolPoint -> MoveTo)

Observer Aborts: Both makes the combat branch interrupt patrol the instant TargetActor is set, and bail out when it's cleared — this is what makes the AI feel reactive.

3. Updating the Blackboard from code (e.g. on seeing the player)

void AEnemyAIController::SetTarget(AActor* Target)
{
    if (UBlackboardComponent* BB = GetBlackboardComponent())
        BB->SetValueAsObject(TEXT("TargetActor"), Target);   // key name must match the BB asset
}
// Clear with BB->ClearValue(TEXT("TargetActor")); to drop back to a lower-priority branch.

Pitfalls

  • AI never starts — the pawn isn't possessed (set the Pawn's Auto Possess AI to "Placed in World or Spawned" and assign the AIController), or RunBehaviorTree was never called.
  • MoveTo instantly fails — no NavMesh in the level (add a Nav Mesh Bounds Volume), or the target is off the navmesh.
  • Branch doesn't react to changes — the gating Decorator's Observer Aborts is set to None; set it to Self/Lower Priority/Both so the tree re-evaluates when the key changes.
  • A task hangs the tree — a custom task returned InProgress and never calls FinishLatentTask. Always complete latent tasks.
  • Blackboard key typos — SetValueAsObject("Taget", ...) silently does nothing; match the key name and type exactly, or use a cached FBlackboardKeySelector.
  • Sequence vs Selector confusion — Sequence = AND (stops on first failure); Selector = OR (stops on first success). Swapping them inverts the behaviour.

References

  • For a custom C++ UBTTaskNode (instant and latent ExecuteTask returning EBTNodeResult, with a FBlackboardKeySelector), read references/custom-bttask.md.
  • Primary docs: "Behavior Trees in Unreal Engine" (https://dev.epicgames.com/documentation/en-us/unreal-engine/behavior-trees-in-unreal-engine).

Related skills

  • game-ai — engine-agnostic AI design (FSM, BT, steering, pathfinding choices).
  • unreal-cpp-gameplay — the AIController and pawn classes in C++.
  • fps-shooter / tower-defense — genres that compose enemy AI.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.