chora

Chora is a flexible architecture for orchestrating chat flows or task execution. It allows you to configure: • Experts — Individual handlers with domain-specific logic. • Expertises — Collections of experts, along with readiness checks or constraints. • Supervisor — A decision-maker that chooses which expertise should handle a given user input. • Orchestrator — The main interface that processes user input, coordinating experts, expertises, and the supervisor.

Features • Modular: You can swap out or add experts, expertises, or supervisors without changing the core orchestrator. • TypeScript Support: Written in TypeScript for type-safe development. • Extensible: Add custom logic for choosing experts, validating input, etc.

Installation

npm install chora or yarn add chora

(Adjust package name if you publish under a different name.)

Basic Usage

Below is a minimal example showing how to set up Chora with one or more experts, an expertise, and a supervisor.

import { Chora, SimpleExpert, SimpleExpertise, SimpleSupervisor } from 'chora'

// 1) Define your experts const mathExpert = new SimpleExpert('MathGuru') const chatExpert = new SimpleExpert('ChatWizard')

// 2) Group experts into expertises // Each expertise can check if it's "ready" to handle the input. const mathExpertise = new SimpleExpertise('MathDomain', [mathExpert]) const chatExpertise = new SimpleExpertise('ChatDomain', [chatExpert])

// 3) Create a supervisor that picks which expertise to use // based on the user input (e.g., first domain that says "I'm ready") const supervisor = new SimpleSupervisor('MainSupervisor')

// 4) Pass these into Chora const chora = new Chora({ expertises: [mathExpertise, chatExpertise], supervisor })

// 5) Process user input const result = chora.processUserInput('Hello, please help me with 2+2') console.log(result) // => [Expert MathGuru] handling: "Hello, please help me with 2+2"

Without a Supervisor

If you’d rather not use a supervisor, you can omit expertises and only provide a direct list of experts. In that case, Chora will pick from the experts array.

import { Chora, SimpleExpert } from 'chora'

const chatExpert = new SimpleExpert('ChatWizard') const mathExpert = new SimpleExpert('MathGuru')

const choraNoSupervisor = new Chora({ experts: [chatExpert, mathExpert] // No supervisor & no expertises })

console.log(choraNoSupervisor.processUserInput('I need chat assistance.')) // => [Expert ChatWizard] handling: "I need chat assistance."

Project Structure

A typical project using Chora might look like this:

my-app/ ├─ src/ │ └─ ... ├─ package.json ├─ tsconfig.json └─ ...

Scripts

All contributions are welcome!

License

MIT

Enjoy orchestrating your chat and task flows with Chora! If you have questions or suggestions, feel free to open an issue or PR.