Skip to content

ONBOARDING.md

Latest commit

 

History

History
90 lines (69 loc) · 2.07 KB

ONBOARDING.md

File metadata and controls

90 lines (69 loc) · 2.07 KB

Onboarding Guide

This guide helps contributors understand the current stream.ai repository without assuming infrastructure that does not exist.

What This Repository Is

Today, this repo is:

  • a small TypeScript library
  • focused on streaming AI responses
  • currently OpenAI-first
  • structured as a single package

It is not currently:

  • a monorepo
  • a multi-package workspace
  • an offline-sync framework

First Things To Read

  1. README.md
  2. ARCHITECTURE.md
  3. docs/openai.md
  4. CONTRIBUTING.md
  5. ROADMAP.md

Local Setup

git clone https://github.com/stream-ai/stream.ai.git
cd stream.ai
npm install
npm run typecheck
npm run test
npm run build

Current Repository Layout

stream.ai/
├── src/
│   ├── builder.ts
│   ├── engine.ts
│   ├── types.ts
│   ├── errors.ts
│   ├── costs.ts
│   ├── edge.ts
│   ├── providers/
│   └── __tests__/
├── README.md
├── ARCHITECTURE.md
├── ROADMAP.md
└── package.json

Where To Start

If you want to understand the codebase quickly:

  1. Read src/types.ts for the public contracts.
  2. Read src/builder.ts for the user-facing API.
  3. Read src/engine.ts for orchestration and aggregation.
  4. Read src/providers/openai.ts for the OpenAI Responses API implementation.
  5. Read docs/openai.md for the provider contract.

Good First Contributions

  • improve examples in the README
  • add or refine tests in src/__tests__
  • improve cost accounting or error handling
  • tighten OpenAI provider behavior
  • add one focused middleware
  • improve OpenAI and package docs

Areas That Need Explicit Discussion First

  • offline storage or synchronization
  • CRDT features
  • package splits or monorepo migration
  • broad architecture changes
  • adding large unvalidated feature claims to docs

Expectations

  • keep scope realistic
  • prefer working software over aspirational docs
  • align documentation with actual implementation
  • favor small, testable changes