Unreal gas
Production-ready Claude Code skills for Unreal Engine 5, llama.cpp, and OpenCode CLI integration. Built by maystudios.
npx -y skills add maystudios/claude-skills --skill unreal-gasAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 13 stars13 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Expert guide for Unreal Engine 5.x Gameplay Ability System (GAS) C++ development. Covers AbilitySystemComponent, GameplayAbilities, GameplayEffects, Attributes/AttributeSets, GameplayTags, GameplayCues, AbilityTasks, prediction/replication, and common patterns. Use when the user asks about GAS, gameplay abilities, gameplay effects, attribute sets, ability system component, gameplay tags in the context of GAS, gameplay cues, ability tasks, or any UAbilitySystemComponent / UGameplayAbility / UGameplayEffect / UAttributeSet related C++ code. Also triggers for questions about ability prediction, ability replication, combo systems using GAS, cooldowns, costs, stacking, or damage/healing pipelines built on GAS.
SKILL.md
6.3 KB, as published. Nobody here has run it
Unreal Engine Gameplay Ability System (GAS) -- C++ Guide
Official Documentation (always consult for latest details)
These are the authoritative sources. Reference them for up-to-date API details, as GAS evolves across engine versions:
The community docs by tranek are the most comprehensive single resource for GAS. They cover advanced topics (prediction internals, optimization, replication modes, pitfalls) that Epic's official docs do not. Always check them for edge cases and production patterns.
The Lyra Sample Project is Epic's recommended working reference implementation for GAS.
Plugin Setup
- Enable Gameplay Abilities plugin in the editor
- Add to
Build.cs:PublicDependencyModuleNames.AddRange(new string[] { "GameplayAbilities", "GameplayTags", "GameplayTasks" }); - UE 5.2 and earlier: call
UAbilitySystemGlobals::Get().InitGlobalData()in your AssetManager or GameInstance. UE 5.3+ does this automatically.
Core Architecture
See references/core-classes.md for the full class reference, setup patterns, and C++ code snippets.
System overview:
Actor
+-- UAbilitySystemComponent (ASC) --- manages everything below
+-- UGameplayAbility instances (granted abilities)
| +-- UAbilityTask instances (async execution)
+-- UAttributeSet instances (numeric data)
+-- Active UGameplayEffect specs (modifiers)
+-- FGameplayTag container (status/state)
+-- GameplayCue triggers (cosmetic FX)
Key relationships:
- ASC is the central hub; one per Actor (place on Pawn or PlayerState)
- Abilities create/apply Effects to modify Attributes and Tags on targets
- Effects auto-trigger Cues when their tags match
GameplayCue.*tags - Tags control ability activation, blocking, and cancellation systemically
- AbilityTasks handle async work (montages, delays, targeting) within abilities
Gameplay Effects Quick Reference
See references/effects-and-attributes.md for detailed GE patterns, modifier math, stacking, cooldowns, and attribute handling.
Duration types:
- Instant -- modifies BaseValue permanently, never tracked as active
- Duration -- modifies CurrentValue, auto-reverts on expiry
- Infinite -- persists until explicitly removed
Modifier aggregation: ((Base + Additive) * Multiplicative) / Division
- Multiply/Divide uses
1 + Sum(Mods - 1)-- two +50% multipliers = +100%, not +125%
Ability Lifecycle
GiveAbility() -> TryActivateAbility() -> CanActivateAbility()
-> ActivateAbility() [override this] -> CommitAbility() [apply cost/cooldown]
-> ... do work (AbilityTasks) ... -> EndAbility()
Four activation methods: explicit handle, GameplayEvent, GameplayEffect tags, Input codes.
Instancing policies:
InstancedPerActor-- recommended default; one instance reused per actorInstancedPerExecution-- new instance each activation; simplest but heaviestNonInstanced-- uses CDO; best performance, C++ only, no state/delegates/RPCs
Networking & Prediction
See references/networking.md for replication modes, prediction details, and multiplayer patterns.
Net Execution Policies: LocalPredicted, LocalOnly, ServerOnly, ServerInitiated.
Key rules:
- ASC replicates Attributes and Tags to all clients, but NOT abilities/effects (bandwidth optimization)
- Non-instant GEs support prediction rollback; instant GEs (damage) do NOT
- Cues use unreliable replication -- cosmetic only, never gameplay logic
- ASC's owning Actor must be locally controlled for remote activation to work
- For PlayerState-based ASC: use Mixed replication for players, Minimal for AI
Common Patterns & Pitfalls
See references/patterns-and-pitfalls.md for implementation recipes and known issues.
Critical pitfalls:
PreAttributeChangeclamping does NOT permanently change modifiers -- clamp BaseValue inPostGameplayEffectExecuteinsteadServer Respects Remote Ability Cancellationcauses more trouble than it's worth -- disable itReplication Policyon GameplayAbility is misleadingly named -- do not use it- PlayerState
NetUpdateFrequencydefaults too low -- increase it or enable Adaptive Network Update Frequency - Removing AttributeSets at runtime can crash clients
- Animation montages must use
PlayMontageAndWaitAbilityTask, not directPlayMontage, for replication