Share your wildest ChatGPT conversations with one click.


ShareGPT is an open-source Chrome Extension for you to share your wildest ChatGPT conversations with one click.


  • Share your ChatGPT conversations with one-click
  • Browse examples on
  • Save your favorite conversations for later
  • Leave comments on conversations

Tech Stack

ShareGPT is built with the following stack:


The ShareGPT API is a REST-styled API that allows you to write and read conversations from our database, exposed as HTTP endpoints.

Conversations Endpoint


You can use this endpoint to add new conversations to our database.

First, if you haven't already, process the ShareGPT conversation using the following code:
function conversationData() {
  const threadContainer = document.querySelector(
    "#__next main div:nth-of-type(1) div:nth-of-type(1) div:nth-of-type(1) div:nth-of-type(1)"
  var result = {
    avatarUrl: getAvatarImage(),
    items: [],
  for (const node of threadContainer.children) {
    const markdownContent = node.querySelector(".markdown");
    // tailwind class indicates human or gpt
    if ([...node.classList].includes("dark:bg-gray-800")) {
        from: "human",
        value: node.textContent,
      // if it's a GPT response, it might contain code blocks
    } else if ([...node.classList].includes("bg-gray-50")) {
        from: "gpt",
        value: markdownContent.outerHTML,
  return result;
function getAvatarImage() {
  // Create a canvas element
  const canvas = document.createElement("canvas");
  const image = document.querySelectorAll("img")[1];
  // Set the canvas size to 30x30 pixels
  canvas.width = 30;
  canvas.height = 30;
  // Draw the img onto the canvas
  canvas.getContext("2d").drawImage(image, 0, 0);
  // Convert the canvas to a base64 string as a JPEG image
  const base64 = canvas.toDataURL("image/jpeg");
  return base64;

Then, send a POST request to the endpoint above with the following payload and request headers:

const res = await fetch("", {
  body: JSON.stringify(conversationData),
  headers: {
    "Content-Type": "application/json",
  method: "POST",

This will return an object with an id attribute which will be the unique identifier for the generated post:

const { id } = await res.json();
const url = `${id}`; // short link to the ShareGPT post


This endpoint takes 3 optional query parameters:

  • type:
    • Used for sorting the results.
    • Takes 2 string values: "new" | "top"
    • "new" sorts conversations by creation time
    • "top" sorts conversations by number of views
    • If undefined, defaults to "top"
  • page:
    • Used for pagination
    • Takes an integer value as a factor of the PAGINATION_LIMIT, which is set to 50.
    • E.g. to get posts 100 - 150, set page to 3
    • If undefined, defaults to 1
  • search
    • Used for filtering records by title.
    • E.g. search = "python" returns all records with the word "python" in the title
    • If undefined, search results are not filtered


await fetch(

This returns a list of conversations with the following type:

interface ConversationMeta {
  id: string; // unique id for the conversation
  title: string; // title of the conversation (first user prompt)
  avatar: string; // base64 encoded URI of the user's avatar
  saves: number; // number of times the conversation is saved on ShareGPT
  comments: number; // number of comments the conversation has on ShareGPT
  views: number; // number of times the conversation has been viewed on ShareGPT
  createdAt: Date; // timestamp when the conversation was creataed