How-To: Drive Node Actions at Runtime

Goal: activate nodes, commit atomic batches, respec, and save/restore loadouts through the manager - all server-validated.

Prerequisites
Quick Start complete: an initialized UCrimsonSkillTreeManager with at least one configured tree.

1. Single action

The simplest path - one node, one action. Routed through a validated Server_ RPC.

cpp
STM->RequestSkillNodeAction(NodeGuid, ECrimsonSkillNodeActionType::Activate);
// Also: Deactivate, IncrementLevel, DecrementLevel

2. Atomic batch (all-or-nothing)

Commit several changes as one transaction - it applies only if every change is valid (great for 'spend N points then confirm' UIs).

cpp
TArray<FSkillNodeChangeRequest> Batch;
Batch.Add({ TalentGuid, ECrimsonSkillNodeActionType::Activate });
Batch.Add({ OtherGuid,  ECrimsonSkillNodeActionType::IncrementLevel });
STM->RequestCommitNodeChanges(Batch);

3. Atomic respec

Wipe a tree and apply a new end-state allocation in one step.

cpp
TArray<FSkillNodeChangeRequest> NewAllocation;
// ... populate with Activate / IncrementLevel requests for the desired end state ...
STM->RequestRespecAndCommit(SkillTreeTag, NewAllocation);

4. Named loadouts

cpp
STM->SaveLoadout(FName("PvP"), SkillTreeTag);
STM->RestoreLoadout(FName("PvP"));  // internally respecs + batch-applies
STM->DeleteLoadout(FName("PvP"));
Authority
All of these are validated server-side. Clients call them freely; the manager fires Server_ RPCs and applies changes only on the server, which replicates the result back. Never mutate node state on a client directly.
Verify
After each call, OnSkillTreeStateUpdated broadcasts. In a Listen Server + Client test, node levels and bIsActive match on both ends.

See also

  • How-To: Save & Load - persist the allocation
  • Concept: Multiplayer - RPCs, FastArrays, authority guards