This repository documents the internal workings and API calls of the ChatGPT web application, aimed at researchers and power users interested in understanding its technical implementation. It provides detailed insights into data structures, API endpoints, and user interactions, offering a transparent view of how ChatGPT operates.

How It Works

The project meticulously dissects the ChatGPT web application's architecture, focusing on its Next.js framework and backend API interactions. It details how user data, session information, and conversation history are managed through various API endpoints like /api/auth/session and /backend-api/conversations. The analysis highlights the use of technologies like Cloudflare for routing and Sentry/Statsig for analytics, providing a granular look at the client-server communication flow.

Quick Start & Requirements

• No installation or execution commands are provided as this is a documentation repository.
• Requires a web browser and familiarity with browser developer tools for verification.
• Links to relevant resources: CONTRIBUTING.md

Highlighted Details

• Detailed breakdown of API requests for actions like sending messages, managing conversations, and user feedback.
• Analysis of data structures returned by various endpoints, including user profiles, model information, and feature settings.
• Documentation of how features like "Custom Instructions," "Code Interpreter," and "Plugins" are managed via API calls.
• Exploration of error handling, including "Too Many Requests" (429) and "Request Entity Too Large" (413) responses.

Maintenance & Community

• Maintained by @tercmd.
• Contributions are welcomed via the CONTRIBUTING.md file.

Licensing & Compatibility

• Licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
• Allows for commercial use and derivative works, provided attribution and share-alike conditions are met.

Limitations & Caveats

The documentation reflects the state of the ChatGPT web app as observed at the time of its creation and may not be entirely up-to-date with subsequent changes. Some API endpoints or behaviors might have evolved.