unrealxu/unrealengine5-skills/ue5-world-interaction
Claude Code and agent skill by unrealxu/unrealengine5-skills
Trust tier 3
- community
Source check pending
Checking that the SKILL.md file is still in the repository.
UE5.6-UE5.8 world interaction systems for pickups, spawners, overlap/trace checks, and visual feedback. Use when requests involve interactive world actors, spawn logic, pickup behavior, interaction radius checks, success/failure feedback, and actor lifecycle control.
Install
npx vskill@latest install unrealxu/unrealengine5-skills/ue5-world-interactionSKILL.md
Quick Start
- Define interaction model: overlap-driven, trace-driven, or explicit use key.
- Define actor set: pickup actor, optional spawner, optional visual mapping data asset.
- Output runtime state transitions from spawn to interaction resolution.
API Anchors (UE5.6-UE5.8)
- Detection anchors:
USphereComponent,UBoxComponentUPrimitiveComponent::OnComponentBeginOverlapUPrimitiveComponent::OnComponentEndOverlapFHitResult- Trace anchors:
UWorld::LineTraceSingleByChannel(...)UWorld::SweepSingleByChannel(...)UKismetSystemLibrary::SphereTraceSingle(...)- Lifecycle anchors:
AActor::SetActorHiddenInGame(...)AActor::SetActorEnableCollision(...)AActor::Destroy(...)AActor::SetActorTickEnabled(...)- Networking anchors:
- server authority validation for interaction success
- replicated result state for client feedback
Interaction Stage Contract
- Every interaction feature must define:
- Detection source (overlap/trace/use-key)
- Eligibility checks (distance, tags, inventory/capacity, authority)
- Success and failure result payloads
- Feedback path (VFX/SFX/UI)
- Post-interaction lifecycle policy (destroy/disable/cooldown/respawn)
- If any item is missing, interaction behavior is underspecified.
Workflow
1) Actor and Component Setup
- Build root + collision + visual components with explicit collision channels.
- Define default states (
active,cooldown,consumed) and replication needs. - Keep collision shape and bounds aligned with intended interaction distance.
2) Detection Path
- Overlap model: bind begin/end overlap delegates on collision component.
- Trace model: run line/sphere traces on input request and validate hit actor/component.
- Keep detection deterministic by explicit channels/profiles.
3) Eligibility and Authority
- Validate target state, range, actor validity, and gameplay conditions.
- In multiplayer, validate success on server before mutating shared state.
- Return structured failure reason on rejection.
4) Resolve Interaction
- On success: apply effect (grant item, trigger state change, start cooldown).
- On failure: keep state stable and emit reason-specific feedback.
- Prevent re-entrant execution during in-progress resolution.
5) Feedback Path
- Emit VFX/SFX/UI feedback for both success and failure paths.
- Separate cosmetic-only effects from authoritative gameplay mutations.
- Keep feedback idempotent for repeated client updates.
6) Lifecycle and Cleanup
- Choose lifecycle explicitly: destroy, hide+disable collision, or reuse via respawn timer.
- If spawner exists, track spawned instances and cleanup on reset/despawn.
- Disable tick for dormant interaction actors when not needed.
Constraints
- Keep interaction validation server-authoritative in multiplayer.
- Ensure collision channels and trace responses are explicit.
- Keep spawner randomization deterministic when reproducibility is needed.
- Avoid hidden side effects in
Tickwithout clear need. - Do not destroy actors before broadcasting required success/failure feedback.
- Keep overlap and trace paths functionally equivalent when both are enabled.
Failure Handling
- Symptom: overlap never triggers.
- Locate: collision enabled state, overlap flags, collision channel responses.
- Fix: enable overlap generation, set proper channel responses, verify component bounds.
- Symptom: trace path misses obvious targets.
- Locate: trace channel/profile and ignore actor list.
- Fix: align trace channel/profile and reduce self/owner ignore misuse.
- Symptom: interaction triggers multiple times.
- Locate: missing in-progress guard and duplicated delegate bindings.
- Fix: add state guard and ensure single bind per component.
- Symptom: client sees success but server rejects.
- Locate: authority checks and RPC validation path.
- Fix: move final validation to server and replicate authoritative result.
- Symptom: consumed pickup remains interactable.
- Locate: lifecycle state and collision disable logic.
- Fix: disable collision/interaction immediately after confirmed success.
- Symptom: spawned interactables leak over time.
- Locate: spawner bookkeeping and reset cleanup path.
- Fix: track spawned instances and destroy/disable stale instances on respawn cycle.
- Symptom: frame spikes near dense interaction areas.
- Locate: per-frame trace/overlap workload and unnecessary ticking.
- Fix: reduce polling frequency, gate traces by input/distance, disable idle tick.
Interaction Authority Ops
- Use server-validated interaction resolution for shared gameplay effects.
- Use client-side prediction only for cosmetic feedback when acceptable.
- Replicate result state, not raw input spam.
- Keep cooldown/timer ownership on authority side for deterministic multiplayer behavior.
UE5.6-UE5.8 Compatibility Notes
- Overlap, trace, and actor lifecycle APIs listed above are stable across UE5.6-UE5.8.
- Prefer explicit collision profile/channel configuration across all supported versions.
Escalation
- Escalate when design requires persistent world state synchronization across sessions.
- Escalate when system must integrate with GAS ability targeting rules.
Security
- Security scanPassed, 100 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/ue5-world-interaction/SKILL.md
Popularity
- Repository stars
- 278
- Repository forks
- 38
- 7-day trend score
- 28
Stars and forks are for the whole GitHub repository, not this skill alone. Refreshed 21 Mar 2026, 7 months ago.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/unrealxu/unrealengine5-skills/ue5-world-interaction)