multiplayer-sync
Skill870 starsChanged 21 days agoMITView on GitHub
Use when synchronizing multiplayer state — MultiplayerSynchronizer, interpolation, prediction, and lag compensation
1 other project uses a version of this.
Adopted by
Most recently taken up by
- decentraland/sdk-skillstheir own version, first seen Oct 25, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- decentraland/sdk-skills★ 4same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository62 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- authoring-godot-prompter-skillsSkill
- releasing-godot-prompterSkill
- 2d-essentialsSkill
- 3d-essentialsSkill
- ability-systemSkill
- addon-developmentSkill
- ai-navigationSkill
- animation-systemSkill
- assets-pipelineSkill
- audio-systemSkill
- beehaveSkill
- camera-systemSkill
- component-systemSkill
- csharp-godotSkill
- csharp-signalsSkill
- dedicated-serverSkill
- dependency-injectionSkill
- dialogue-managerSkill
- dialogue-systemSkill
- event-busSkill
- export-pipelineSkill
- gdextensionSkill
- gdscript-advancedSkill
- gdscript-patternsSkill
- godot-brainstormingSkill
- godot-code-reviewSkill
- godot-debuggingSkill
- godot-grillSkill
- godot-mentorSkill
- godot-optimizationSkill
- godot-project-setupSkill
- godot-testingSkill
- godot-uiSkill
- hud-systemSkill
- input-handlingSkill
- inventory-systemSkill
- limboaiSkill
- localizationSkill
- math-essentialsSkill
- mobile-developmentSkill
- multiplayer-basicsSkill
- multithreadingSkill
- particles-vfxSkill
- phantom-cameraSkill
- physics-systemSkill
- player-controllerSkill
- popochiuSkill
- procedural-generationSkill
- rendering-pipelineSkill
- resource-patternSkill
- responsive-uiSkill
- save-loadSkill
- scene-organizationSkill
- shader-basicsSkill
- state-machineSkill
- tween-animationSkill
- using-godot-prompterSkill
- web-exportSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
All examples target Godot 4.3+ with no deprecated APIs.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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.
Files
Multiplayer Synchronization in Godot 4.3+
All examples target Godot 4.3+ with no deprecated APIs. GDScript is shown first, then C#.
Related skills: multiplayer-basics for ENet setup, RPCs, and authority model, dedicated-server for headless export and deployment, physics-system for physics interpolation and RigidBody synchronization.
1. MultiplayerSynchronizer
MultiplayerSynchronizer is Godot's built-in node for replicating properties across the network. Add it as a child of the node whose state you want to share.
What It Does
- Sends property values from the authority peer to all others at a configured interval
- Supports both delta sync ("On Change" properties, sent reliably when changed) and full sync ("Always" properties, sent unreliably every interval)
- Allows visibility filters to control which peers receive updates
Replication Config in the Editor
- Select the
MultiplayerSynchronizernode in the scene tree. - In the Inspector, open Replication and click Add Property.
- Pick the parent node path and property name (e.g.
position,velocity). - Set Sync (send every interval) or Spawn (send only on spawn) per property.
- Set the Replication Interval (seconds).
0means every network process frame (the multiplayer API polls once per process frame, not per physics tick).
Key Properties
| Property | Description |
replication_interval | Seconds between full sync updates. 0 (default) = every network process frame |
delta_interval | Seconds between delta sync updates. 0 (default) = every network process frame |
public_visibility | When true, updates go to all peers (default) |
add_visibility_filter(callable) | Method, not a property: adds a Callable that takes a peer ID and returns true if that peer should receive updates |
Delta vs Full Sync
| Mode | How It Works | Best For |
| Full sync | Sends every "Always" property every replication_interval, unreliably | Continuously changing values (position, velocity) |
| Delta sync | Sends the "On Change" properties that changed, every delta_interval, reliably | Values that change rarely but must arrive (health, state) |
Each property uses one mode, so the two intervals govern different property sets — replication_interval is not a full-state refresh of the "On Change" properties.
Visibility Filters (GDScript)
# Only send updates to peers within 500 units of this object.
func _ready() -> void:
$MultiplayerSynchronizer.add_visibility_filter(_is_peer_in_range)
func _is_peer_in_range(peer_id: int) -> bool:
var peer_player := _get_player_node(peer_id)
if peer_player == null:
return false
return global_position.distance_to(peer_player.global_position) <= 500.0Visibility Filters (C#)
// Only send updates to peers within 500 units of this object.
public override void _Ready()
{
var sync = GetNode<MultiplayerSynchronizer>("MultiplayerSynchronizer");
sync.AddVisibilityFilter(Callable.From<int, bool>(IsPeerInRange));
}
private bool IsPeerInRange(int peerId)
{
var peerPlayer = GetPlayerNode(peerId);
if (peerPlayer is null)
return false;
return GlobalPosition.DistanceTo(peerPlayer.GlobalPosition) <= 500.0f;
}2. Property Synchronization
What to Sync
Sync the minimal state needed to reconstruct the visual on remote peers. Typical properties:
| Property | Type | Notes |
position | Vector2 / Vector3 | Core transform — sync every frame or use interpolation |
velocity | Vector2 / Vector3 | Helps remote prediction stay ahead of position snaps |
health | int / float | Sync reliably on change; delta sync is ideal |
animation_state | String / int | Sync on change; use an enum int to save bandwidth |
is_crouching | bool | Low-change boolean; delta sync or RPC on change |
Synced Player (GDScript)
# synced_player.gd
extends CharacterBody2D
## Sync interval in seconds — exposed so designers can tune per object type.
@export var sync_interval: float = 0.05 # 20 Hz
@export var speed: float = 200.0
# These properties are listed in the MultiplayerSynchronizer replication config.
var synced_position: Vector2 = Vector2.ZERO
var synced_velocity: Vector2 = Vector2.ZERO
var synced_health: int = 100
var synced_anim: int = 0 # 0 = idle, 1 = run, 2 = jump
@onready var _sync: MultiplayerSynchronizer = $MultiplayerSynchronizer
func _ready() -> void:
_sync.replication_interval = sync_interval
# Only the authority (owner) drives movement.
set_physics_process(is_multiplayer_authority())
func _physics_process(_delta: float) -> void:
# Authority: write canonical state so MultiplayerSynchronizer can replicate it.
synced_position = global_position
synced_velocity = velocity
synced_anim = _compute_anim_state()Synced Player (C#)
// SyncedPlayer.cs
using Godot;
public partial class SyncedPlayer : CharacterBody2D
{
/// <summary>Sync interval in seconds. Exposed so designers can tune per object type.</summary>
[Export] public float SyncInterval { get; set; } = 0.05f; // 20 Hz
[Export] public float Speed { get; set; } = 200.0f;
// These properties are listed in the MultiplayerSynchronizer replication config.
public Vector2 SyncedPosition { get; set; } = Vector2.Zero;
public Vector2 SyncedVelocity { get; set; } = Vector2.Zero;
public int SyncedHealth { get; set; } = 100;
public int SyncedAnim { get; set; } = 0; // 0=idle, 1=run, 2=jump
private MultiplayerSynchronizer _sync = null!;
public override void _Ready()
{
_sync = GetNode<MultiplayerSynchronizer>("MultiplayerSynchronizer");
_sync.ReplicationInterval = SyncInterval;
SetPhysicsProcess(IsMultiplayerAuthority());
}
public override void _PhysicsProcess(double delta)
{
// Authority: write canonical state for replication.
SyncedPosition = GlobalPosition;
SyncedVelocity = Velocity;
SyncedAnim = ComputeAnimState();
}
private int ComputeAnimState()
{
if (!IsOnFloor()) return 2;
return Velocity.Length() > 1f ? 1 : 0;
}
}3. Interpolation
MultiplayerSynchronizer updates target properties at sync intervals (e.g., 30 Hz), but rendering runs at frame rate (60+ Hz). Without interpolation, remote players appear to teleport between snapshots. The fix: on each synchronized signal keep the previous and the new position, and lerp between them in _process by the time elapsed since the update divided by the sync interval; a timestamped snapshot buffer (render ~100 ms behind) is the more robust version.
See references/interpolation.md for the full GDScript and C# previous/current-snapshot interpolation (
synchronizedsignal, elapsed-time lerp in_process).
4. Client-Side Prediction
For local-player responsiveness: predict movement immediately on client, send input to server, reconcile when server snapshot arrives. If server diverges from client prediction beyond a threshold, snap; otherwise smoothly lerp the correction over 100-200 ms.
See references/client-prediction.md for the full predict-and-reconcile pattern (input ring buffer, server reconciliation, replay) in GDScript + C#.
5. Lag Compensation
For hit-scan weapons in fast-paced games: when the server validates a hit, it rewinds the world state to the client's view-time (now - client_rtt/2 - interp_delay) and tests the hit against that historical state.
See references/lag-compensation.md for the snapshot-history pattern, view-time calculation, and a hit-scan validator in GDScript + C#.
6. State vs Input Synchronization
Choose the synchronization model that fits your game's needs:
| Factor | Sync State | Sync Inputs |
| What is sent | Current property values (position, health, etc.) | Player input actions each frame |
| Who simulates | Authority only; others receive results | All peers run the same simulation |
| Determinism required | No | Yes — every peer must produce identical output from the same inputs |
| Bandwidth | Higher — full state sent each interval | Lower — small input structs per frame |
| Responsiveness | Lower — non-authority peers wait for next sync tick | Higher — local prediction is trivial when deterministic |
| Complexity | Lower — no reconciliation loop | Higher — requires deterministic physics, fixed-point math, or lockstep |
| Best for | Action games, shooters, most real-time games | Fighting games, RTS, turn-based, simulation games |
| Lag compensation needed | Yes, for hit detection | Usually not — all peers are in sync |
Hybrid approach (most real-time games): sync inputs for the local player's character (enabling prediction), sync state for all other objects and game events.
7. Bandwidth Optimization
Four levers: sync only changed properties (replication-config flag per property), quantize floats (components as 16-bit centimetres instead of floats — RPC arguments and synced properties compress ints to 1/2/4/8 bytes plus a type byte, so a 16-bit int costs 3 bytes against 8 for a float), distance-based sync rate (far-away objects sync at 5 Hz, close at 30 Hz), and channel selection (reliable for state changes that must arrive, unreliable for position streams that get superseded).
See references/bandwidth-optimization.md for full GDScript + C# recipes for each lever, plus the reliable-vs-unreliable channel decision tree.
8. Implementation Checklist
MultiplayerSynchronizeris a direct child of the node it replicates- Only the authority peer writes to synced properties; others are read-only
set_multiplayer_authority()is called at spawn time with the correct peer IDreplication_intervalanddelta_intervalare tuned for the object's update rate- Remote player visuals use interpolation in
_process, not_physics_process - Interpolation stores previous and current state and blends by the time since the last update (not
Engine.get_physics_interpolation_fraction(), which tracks physics ticks, not network updates) - Client-side prediction is applied only to the local player's own character
- Pending input buffer is bounded (max ~128 ticks) to prevent memory growth
- Reconciliation threshold prevents jitter from micro-corrections
- Position and velocity use
unreliableRPC; state changes usereliable - Float quantization is applied before sending position data over the network
- Lag compensation snapshot history is pruned each tick to a bounded window
- Server validates all hit detection; clients never self-report kills
- Distance-based sync rate reduces bandwidth for far-away objects
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
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.

