> ## Documentation Index
> Fetch the complete documentation index at: https://docs.htmldrop.app/llms.txt
> Use this file to discover all available pages before exploring further.

# ChatGPT

> Add htmldrop as a custom connector, then publish from any conversation.

ChatGPT can talk to htmldrop through a custom MCP connector. Connectors cannot be
installed from inside a chat — that is deliberate on OpenAI's side, since it is
exactly what a prompt-injection attack would try — so you add it once in
settings.

<Steps>
  <Step title="Turn on developer mode">
    **Settings → Apps → Advanced** and enable **Developer mode**. On Business and
    Enterprise workspaces an admin may need to permit custom connectors first.
  </Step>

  <Step title="Create the connector">
    **Settings → Connectors → Create**, then enter:

    ```
    https://htmldrop.app/mcp
    ```

    Give it a name and a description — the model reads the description when
    deciding whether to use it, so something like "Publishes HTML or Markdown to
    a live URL" works well.
  </Step>

  <Step title="Approve the sign-in">
    The first publish opens a browser window for OAuth. Approve once; ChatGPT
    holds and refreshes the token from then on.
  </Step>
</Steps>

## Using it

> Make me a one-page portfolio and publish it with htmldrop

ChatGPT writes the HTML, calls `htmldrop_publish`, and hands back the live URL.

## Without a connector

If you would rather not add one, ask ChatGPT for a **single self-contained HTML
file**, download it, and drag it onto [htmldrop.app](https://htmldrop.app). Same
result, one extra step.