KS
Killer-Skills

koota — how to use koota how to use koota, koota vs redux, koota install guide, what is koota, koota react integration, koota typescript support, koota state management, koota entity component system

v1.0.0
GitHub

About this Skill

Ideal for React and TypeScript Agents seeking optimized state management with composable traits and entity-based data handling. Koota is a state management system that uses entities with composable traits, relations, and archetypes to manage data in React and TypeScript applications.

Features

Manages state using entities with composable traits
Supports schema-based (SoA) and callback-based (AoS) trait definitions
Enables directional connections between entities via relations
Utilizes archetypes for unique combinations of traits
Provides a world context for all entities and their data

# Core Topics

pmndrs pmndrs
[0]
[0]
Updated: 3/7/2026

Quality Score

Top 5%
42
Excellent
Based on code quality & docs
Installation
SYS Universal Install (Auto-Detect)
Cursor IDE Windsurf IDE VS Code IDE
> npx killer-skills add pmndrs/koota/references/react-patterns.md

Agent Capability Analysis

The koota MCP Server by pmndrs is an open-source Categories.community integration for Claude and other AI agents, enabling seamless task automation and capability expansion. Optimized for how to use koota, koota vs redux, koota install guide.

Ideal Agent Persona

Ideal for React and TypeScript Agents seeking optimized state management with composable traits and entity-based data handling.

Core Value

Empowers agents to efficiently manage state using entities with composable traits, supporting schema-based, callback-based, and tag-based data definitions with Relations and Archetypes, leveraging the Koota ECS for real-time state updates.

Capabilities Granted for koota MCP Server

Managing complex application states with composable traits
Building entity-based data models with Relations and Archetypes
Optimizing real-time state updates in React applications

! Prerequisites & Limits

  • Requires React and TypeScript environments
  • Limited to entity-based state management
Project
SKILL.md
8.1 KB
.cursorrules
1.2 KB
package.json
240 B
Ready
UTF-8

# Tags

[No tags]
SKILL.md
Readonly

Koota ECS

Koota manages state using entities with composable traits.

Glossary

  • Entity - A unique identifier pointing to data defined by traits. Spawned from a world.
  • Trait - A reusable data definition. Can be schema-based (SoA), callback-based (AoS), or a tag.
  • Relation - A directional connection between entities to build graphs.
  • World - The context for all entities and their data (traits).
  • Archetype - A unique combination of traits that entities share.
  • Query - Fetches entities matching an archetype. The primary way to batch update state.
  • Action - A discrete, synchronous data mutation (create, update, destroy). Reusable from any call site.
  • System - A reactive orchestrator that observes state changes and coordinates work, including async workflows. Runs in the frame loop or event callbacks.

Design Principles

Data-oriented

Behavior is separated from data. Data is defined as traits, entities compose traits, and systems mutate data on traits via queries. See Basic usage for a complete example.

Composable systems

Design systems as small, single-purpose units rather than monolithic functions that do everything in sequence. Each system should handle one concern so that behaviors can be toggled on/off independently.

typescript
1// Good: Composable systems - each can be enabled/disabled independently 2function applyVelocity(world: World) {} 3function applyGravity(world: World) {} 4function applyFriction(world: World) {} 5function syncToDOM(world: World) {} 6 7// Bad: Monolithic system - can't disable gravity without disabling everything 8function updatePhysicsAndRender(world: World) { 9 // velocity, gravity, friction, DOM sync all in one function 10}

This enables feature flags, debugging (disable one system to isolate issues), and flexible runtime configurations.

Decouple view from logic

Separate core state and logic (the "core") from the view ("app"):

  • Run logic independent of rendering
  • Swap views while keeping state (2D ↔ 3D)
  • Run logic in a worker or on a server

Prefer traits + actions over classes

Prefer not to use classes to encapsulate data and behavior. Use traits for data and actions for behavior. Only use classes when required by external libraries (e.g., THREE.js objects) or the user prefers it.

Directory structure

If the user has a preferred structure, follow it. Otherwise, use this guidance: the directory structure should mirror how the app's data model is organized. Separate core state/logic from the view layer:

  • Core - Pure TypeScript. Traits, systems, actions, world. No view imports.
  • View - Reads from world, mutates via actions. Organized by domain/feature.
src/
├── core/              # Pure TypeScript, no view imports
│   ├── traits/
│   ├── systems/
│   ├── actions/
│   └── world.ts
└── features/          # View layer, organized by domain

Files are organized by role, not by feature slice. Traits and systems are composable and don't map cleanly to features.

For detailed patterns and monorepo structures, see references/architecture.md.

Trait types

TypeSyntaxUse whenExamples
SoA (Schema)trait({ x: 0 })Simple primitive dataPosition, Velocity, Health
AoS (Callback)trait(() => new Thing())Complex objects/instancesRef (DOM), Keyboard (Set)
Tagtrait()No data, just a flagIsPlayer, IsEnemy, IsDead

Trait naming conventions

TypePatternExamples
TagsStart with IsIsPlayer, IsEnemy, IsDead
RelationsPrepositionalChildOf, HeldBy, Contains
TraitNounPosition, Velocity, Health

Relations

Relations build graphs between entities such as hierarchies, inventories, targeting.

typescript
1import { relation } from 'koota' 2 3const ChildOf = relation({ autoDestroy: 'orphan' }) // Hierarchy 4const Contains = relation({ store: { amount: 0 } }) // With data 5const Targeting = relation({ exclusive: true }) // One target only 6 7// Build graph 8const parent = world.spawn() 9const child = world.spawn(ChildOf(parent)) 10 11// Query children of parent 12const children = world.query(ChildOf(parent)) 13 14// Query all entities with any ChildOf relation 15const allChildren = world.query(ChildOf('*')) 16 17// Get targets from entity 18const items = entity.targetsFor(Contains) // Entity[] 19const target = entity.targetFor(Targeting) // Entity | undefined

For detailed patterns, traversal, ordered relations, and anti-patterns, see references/relations.md.

Basic usage

typescript
1import { trait, createWorld } from 'koota' 2 3// 1. Define traits 4const Position = trait({ x: 0, y: 0 }) 5const Velocity = trait({ x: 0, y: 0 }) 6const IsPlayer = trait() 7 8// 2. Create world and spawn entities 9const world = createWorld() 10const player = world.spawn(Position({ x: 100, y: 50 }), Velocity, IsPlayer) 11 12// 3. Query and update 13world.query(Position, Velocity).updateEach(([pos, vel]) => { 14 pos.x += vel.x 15 pos.y += vel.y 16})

Entities

Entities are unique identifiers that compose traits. Spawned from a world.

typescript
1// Spawn 2const entity = world.spawn(Position, Velocity) 3 4// Read/write traits 5entity.get(Position) // Read trait data 6entity.set(Position, { x: 10 }) // Write (triggers change events) 7entity.add(IsPlayer) // Add trait 8entity.remove(Velocity) // Remove trait 9entity.has(Position) // Check if has trait 10 11// Destroy 12entity.destroy()

Entity IDs

An entity is internally a number packed with entity ID, generation ID (for recycling), and world ID. Safe to store directly for persistence or networking.

typescript
1entity.id() // Just the entity ID (reused after destroy) 2entity // Full packed number (unique forever)

Typing

Use TraitRecord to get the type that entity.get() returns

typescript
1type PositionRecord = TraitRecord<typeof Position>

Queries

Queries fetch entities matching an archetype and are the primary way to batch update state.

typescript
1// Query and update 2world.query(Position, Velocity).updateEach(([pos, vel]) => { 3 pos.x += vel.x 4 pos.y += vel.y 5}) 6 7// Read-only iteration (no write-back) 8const data: Array<{ x: number; y: number }> = [] 9world.query(Position, Velocity).readEach(([pos, vel]) => { 10 data.push({ x: pos.x, y: pos.y }) 11}) 12 13// Get first match 14const player = world.queryFirst(IsPlayer, Position) 15 16// Filter with modifiers 17world.query(Position, Not(Velocity)) // Has Position but not Velocity 18world.query(Or(IsPlayer, IsEnemy)) // Has either trait

Note: updateEach/readEach only return data-bearing traits (SoA/AoS). Tags, Not(), and relation filters are excluded:

typescript
1world.query(IsPlayer, Position, Velocity).updateEach(([pos, vel]) => { 2 // Array has 2 elements - IsPlayer (tag) excluded 3})

For tracking changes, caching queries, and advanced patterns, see references/queries.md.

React integration

Imports: Core types (World, Entity) from 'koota'. React hooks from 'koota/react'.

Change detection: entity.set() and world.set() trigger change events that cause hooks like useTrait to rerender. For AoS traits where you mutate objects directly, manually signal with entity.changed(Trait).

For React hooks and actions, see references/react-hooks.md.

For component patterns (App, Startup, Renderer, view sync, input), see references/react-patterns.md.

Runtime

Systems query the world and update entities. Run them via frameloop (continuous) or event handlers (discrete).

For systems, frameloop, event-driven patterns, and time management, see references/runtime.md.

Related Skills

Looking for an alternative to koota or building a Categories.community AI Agent? Explore these related open-source MCP Servers.

View All

widget-generator

Logo of f
f

widget-generator is an open-source AI agent skill for creating widget plugins that are injected into prompt feeds on prompts.chat. It supports two rendering modes: standard prompt widgets using default PromptCard styling and custom render widgets built as full React components.

149.6k
0
Design

chat-sdk

Logo of lobehub
lobehub

chat-sdk is a unified TypeScript SDK for building chat bots across multiple platforms, providing a single interface for deploying bot logic.

73.0k
0
Communication

zustand

Logo of lobehub
lobehub

The ultimate space for work and life — to find, build, and collaborate with agent teammates that grow with you. We are taking agent harness to the next level — enabling multi-agent collaboration, effortless agent team design, and introducing agents as the unit of work interaction.

72.8k
0
Communication

data-fetching

Logo of lobehub
lobehub

The ultimate space for work and life — to find, build, and collaborate with agent teammates that grow with you. We are taking agent harness to the next level — enabling multi-agent collaboration, effortless agent team design, and introducing agents as the unit of work interaction.

72.8k
0
Communication