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

# Gemini Images connector

> Generate, edit, and compose images with Google Gemini and Imagen models from AgentRuntime workflows.

The **geminiimage** MCP adapter connects to [Google AI Studio](https://aistudio.google.com/) image APIs. Generate images from prompts, edit existing images with instructions, and compose multiple inputs — using Gemini multimodal models or Imagen predict endpoints.

## Prerequisites

* A Google AI Studio account with an **API key**
* Billing or quota enabled for Gemini image and/or Imagen models on your key
* For edit/compose: source images as base64 (data URL prefixes accepted)
* **project\_contributor** access in AgentRuntime

## Connect in AgentRuntime

<Steps>
  <Step title="Get a Google AI Studio API key">
    Create a key at [aistudio.google.com/apikey](https://aistudio.google.com/apikey). Enable the image models you plan to use.
  </Step>

  <Step title="Create a connection">
    Go to **Connections** and click **New custom connection** to create an API key connection with `APIKey`. Set `DefaultModel` (e.g. `gemini-2.5-flash-image`) if desired.
  </Step>

  <Step title="Add geminiimage MCP instance">
    Go to **MCP**, click **Add instance** for **geminiimage**, wire your connection on **Instance config**, set the profile **active**, and save.
  </Step>

  <Step title="Generate a test image">
    Run **mcp\_call** → `gemini_text_to_image` with a simple prompt to verify model access.
  </Step>
</Steps>

See [Authentication](/integrations/authentication) for binding connections to MCP instances.

<Note>
  `gemini_edit_image` and `gemini_compose_images` require Gemini image models. Imagen models (`imagen-4-*`, `imagen-3-*`) support text-to-image via predict but not edit/compose handlers.
</Note>

## What you can build

* **Brand assets** — `gemini_text_to_image` with Imagen aspect ratios for ads and social formats.
* **Photo retouching** — Product photo base64 → `gemini_edit_image` with "remove background" or style prompts.
* **Collage generation** — Multiple reference images → `gemini_compose_images` with layout instructions.
* **Multimodal reports** — LLM writes caption → image tool → append to Google Slides or Docs.

## Tools

| Tool                    | Description                                                     |
| ----------------------- | --------------------------------------------------------------- |
| `gemini_text_to_image`  | Generate from prompt (Gemini generateContent or Imagen predict) |
| `gemini_edit_image`     | Edit a base64 image with a text prompt (Gemini models only)     |
| `gemini_compose_images` | Combine multiple input images per prompt (Gemini models only)   |

## Example

Generate a square marketing image with Imagen:

```json theme={null}
{
  "id": "gemini-image",
  "type": "mcp_call",
  "name": "Generate product image",
  "tool_name": "gemini_text_to_image",
  "tool_args": {
    "prompt": "Minimal flat illustration of a cloud workflow, brand colors blue and white",
    "model": "imagen-4.0-generate-001",
    "aspect_ratio": "1:1",
    "number_of_images": 1
  },
  "timeout_s": 120
}
```

## Configuration

| Key            | Required | Description                                                     |
| -------------- | -------- | --------------------------------------------------------------- |
| `APIKey`       | Yes      | Google AI Studio API key                                        |
| `DefaultModel` | No       | Default when tools omit `model` (e.g. `gemini-2.5-flash-image`) |
| `APIBase`      | No       | Gemini API base URL                                             |

## Troubleshooting

| Issue                   | Fix                                                                              |
| ----------------------- | -------------------------------------------------------------------------------- |
| Imagen rejected on edit | Use Gemini model for `gemini_edit_image` and `gemini_compose_images`             |
| Empty image in response | Check safety filters; simplify prompt; verify model ID spelling                  |
| Wrong aspect ratio      | Set `aspect_ratio` for Imagen; Gemini generateContent ignores Imagen-only fields |
| Quota errors            | Review AI Studio quotas; reduce concurrent workflow image steps                  |

## Related

* [Integrations quickstart](/integrations/quickstart)
* [OpenAI Images connector](/connectors/openaiimage) — alternative image generation
* [Connector catalog](/integrations/connector-catalog)
* [Troubleshooting](/platform/troubleshooting)