
This official template cannot be run directly. Create an editable duplicate, restore your ChatGPT login, and publish your personal image Bot.
This template cannot be run directly. Select Create Editable Duplicate to save a personal copy. In its Browser settings, select Standard Browser and configure Website login for your ChatGPT account. Publish your copy if prompted, then run or connect that personal Bot.

Turn your own logged-in ChatGPT Image 2.5 subscription into a reusable BrowserAct Bot. Submit an image prompt and optional reference images, then receive a downloadable PNG result for image generation or editing.
This is not the official OpenAI API. The copied Bot automates the ChatGPT web experience through your own authenticated browser profile and can be invoked through the BrowserAct API or MCP after you publish it.
Parameter | Configuration |
|---|---|
| String · Required |
| Image list · Optional |
| Integer · Required |
| Integer · Required |
| Integer · Required |
The table shows one real result from a completed image-editing run.
Field | Result |
|---|---|
|
|
|
|
|
|
|
|
|
|
| Replace the background with a dark studio scene while preserving the subject. |
Select Create Editable Duplicate at the top right of this template page. Save the copy to your own BrowserAct workspace and open that Bot. Configure login on your personal copy, not on the shared template.
Open your copied Bot's Browser settings. Under Browser profile, select Standard Browser before signing in. Standard Browser saves successful login updates for future runs; Rotating Browser does not keep its temporary session. ChatGPT can still expire a session or request verification.
In Browser settings > Website login, choose one of three login methods: Account credentials, Import cookies, or Log in manually. Follow the Website Login guide to set up your preferred method.

Choose a login method in your copied Bot's Browser settings.
Publish your personal copy if prompted, then open Run. Enter your image prompt and add reference_images when needed. Run your personal Bot to generate or edit an image. Use the same personal Bot for later runs and Agent connections.
If a run asks you to log in, return to Browser settings > Website login on the same personal Bot, refresh your login using the Website Login guide, and start a new run.

An editable duplicate can generate a different MCP Tool Call Name and different input descriptions. Review the copied Bot's MCP configuration and adapt it using the original tested Bot examples below. These are reference values, not fields that must remain identical in every copy.
Authorization: Bearer <BROWSERACT_MCP_KEY>. After the Agent discovers the tool, ask it to generate or edit an image in natural language.Use these descriptions as a starting point when editing the copied Bot. The original tested Bot uses AI-inferred for all five tool inputs.
MCP field | Original Bot example | Mode |
|---|---|---|
|
| Unique name |
| Creates an image on ChatGPT Images from a text prompt, optionally guided by reference images. Pass the literal | Tool level |
| The exact image-generation prompt; it is not summarized, translated, rewritten, reordered, or split. | AI-inferred |
| Required. Pass | AI-inferred |
| Maximum prompt characters; a longer prompt is cut back to its last complete sentence. | AI-inferred |
| Maximum number of reference images accepted from | AI-inferred |
| Maximum size in MB accepted for a single reference image. Exceeding it fails the run; images are never skipped or downscaled silently. | AI-inferred |
The screenshot below shows a copied Bot whose generated Tool Call Name and descriptions need to be reviewed and adjusted against the original Bot examples above before the tool is exposed to an MCP Server.

See the complete BrowserAct MCP setup guide.
Need help restoring the login session or connecting your copied Bot?