opencoredev/login-with-chatgpt

A simple SDK that lets your users log in with their ChatGPT account.

What it solves

Most apps that want to use ChatGPT models either require building their own auth or forcing users to get separate API keys. This SDK solves that by letting your users log in with their own ChatGPT account, bringing their subscription to your app. Tokens never touch the browser — they stay behind a proxy, so you avoid security risks from exposing credentials.

How it works

The SDK provides a handler that keeps tokens behind a proxy path. When a user clicks the login button, the browser gets a session cookie, then your backend asks which models the user's account has access to and streams responses from there. The whole flow is split into packages: core handles OAuth, token refresh, and model discovery; server provides the backend endpoints; react gives you the button and hook; and the AI package provides providers for the Vercel AI SDK.

Who it's for

Developers building web apps with Node.js 18+ who want to offer ChatGPT-powered features without handling API keys themselves, and who want users to use their own ChatGPT subscriptions. It's also useful if you need image generation, custom tool handoffs, or experimental voice-based interaction.

Highlights

  • Users bring their own ChatGPT subscription, so you don't need to manage API keys
  • Tokens are protected via HttpOnly cookies and a proxy handler; they never reach the browser
  • Works directly with the Vercel AI SDK: you can call streamText() from the client safely
  • Supports image generation and editing (size, quality, format, masks, streaming previews)
  • Experimental GPT Live WebRTC audio with native interruption
  • Desktop-style app-server handoffs for allowlisted tools with explicit confirmations
  • Open source under the MIT license
  • Includes an agent skill (for Claude Code, Cursor, or Codex) that correctly wires the SDK without inventing API keys or model slugs

Related

  • Project
    berelevant-ai/slopless

    Slopless is a Markdown linter that flags vague, clichéd, or overconfident English prose. It runs deterministic rules locally by default, offers optional AI-based semantic checks via TypeSafe, and can be used as a CLI, a JavaScript library, or a textlint preset. It also installs agent skills for Codex and Claude Code, and ships with SLSA provenance attestations.

  • Project
    YUHAI0/fin-agent

    An AI-powered financial analysis assistant that uses LLMs and Tushare data to enable natural language stock screening, strategy backtesting, and portfolio monitoring.

  • Project
    anthropics/anthropic-sdk-csharp

    The Claude SDK for C# is the official .NET library for calling the Claude API. Install via `dotnet add package Anthropic`, create an `AnthropicClient`, build a `MessageCreateParams` with model, max tokens, and messages, then call `client.Messages.Create()`. Requires .NET Standard 2.0+. Licensed under Apache 2.0 (older versions MIT). Note: pre-v10 community builds moved to `tryAGI.Anthropic`.

  • Project
    hikariming/dshfind

    dshfind is a community site for DeepSeek Harness (DSH) that offers structured lessons on DSH principles, a plugin marketplace that automatically aggregates GitHub repositories tagged `dsh-plugin`, and best-practice guides with community rankings. It solves discovery and education gaps for the DSH ecosystem.