# Amazon Bedrock Nova Canvas [Health: Active]

**Category:** 🛠️ Other Tools and Integrations  
**Repository:** https://github.com/zxkane/mcp-server-amazon-bedrock  
**GitHub Stars:** 25  
**npm Downloads (last month):** 82  
**Views:** 4  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/amazon-bedrock-nova-canvas

## Description
Use Amazon Nova Canvas model for image generation.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "amazon-bedrock-nova-canvas": {
    "command": "npx",
    "args": ["-y","@zxkane/mcp-server-amazon-bedrock"],
    "env": {
      "AWS_ACCESS_KEY_ID": "",
      "AWS_SECRET_ACCESS_KEY": "",
      "AWS_REGION": "",
      "AWS_PROFILE": ""
    }
  }
}
```

**Requires environment variables:** `AWS_ACCESS_KEY_ID`, `AWS_SECRET_ACCESS_KEY`, `AWS_REGION`, `AWS_PROFILE` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What Amazon Bedrock Nova Canvas MCP server does

Amazon Bedrock Nova Canvas MCP server exposes Amazon Bedrock's Nova Canvas model through the Model Context Protocol. It provides one documented tool, `generate_image`, for creating images from text descriptions. The tool can also exclude specified elements, control output dimensions and quality, set prompt-adherence strength, choose a generation seed, and request multiple images in one call.

The server fits workflows where an MCP-connected assistant needs to turn descriptions into images without a separate application-specific image-generation integration. An AWS account, access to the Nova Canvas model, suitable Bedrock permissions, and Node.js 18 or later are required.

## How it works

An MCP client starts the package with `npx`, using the command and arguments shown in the Claude Desktop configuration example. The server then uses the configured AWS authentication method to call Amazon Bedrock in the selected region. AWS credentials may come from environment variables, the standard AWS credentials file, or an IAM role when the server runs on AWS infrastructure.

The `generate_image` input requires a `prompt` between 1 and 1,024 characters. Optional settings include `negativePrompt`, `width`, `height`, `quality`, `cfg_scale`, `seed`, and `numberOfImages`. Width and height default to 1,024 pixels, quality defaults to `standard`, and one image is generated by default. The documented ranges include 1.1–10 for `cfg_scale`, 0–858,993,459 for `seed`, and 1–5 for `numberOfImages`.

For more predictable prompting, put unwanted elements in `negativePrompt` instead of expressing them with terms such as “no” or “without” in the main prompt. Resolution, image count, and quality can affect generation time and may create timeout concerns in the calling implementation.

## Setup and configuration

Install and build the project locally with `npm install` followed by `npm run build`, or configure an MCP client to run the published package with `npx -y @zxkane/mcp-server-amazon-bedrock`.

AWS configuration can use these environment variables:

- `AWS_ACCESS_KEY_ID` and `AWS_SECRET_ACCESS_KEY` for direct credentials
- `AWS_REGION` to select the Bedrock region; the example states that it defaults to `us-east-1`
- `AWS_PROFILE` to select a named profile from the AWS credentials file

Direct access-key variables are optional when using a credentials file or an IAM role. The AWS identity still needs the permissions required to access Amazon Bedrock and Nova Canvas.

## Tools and capabilities

Amazon Bedrock Nova Canvas MCP server documents the following tool:

- `generate_image`: Generates one or more images from a text prompt.
- Negative prompts: Specifies elements to leave out.
- Output controls: Sets width, height, standard or premium quality, and prompt-adherence strength.
- Reproducibility: Accepts a seed value for deterministic generation.
- Batching: Requests between one and five images per call.

## Limitations and notes

Successful use depends on AWS account access, model availability, credentials, and permissions. Higher resolution, premium quality, or larger batches can increase generation time and may require timeout handling in the client or application.

The project is licensed under MIT. The README identifies Claude Desktop as an integration target and does not document other MCP clients.

_Full upstream README: https://allmcps.com/mcp/amazon-bedrock-nova-canvas/readme_