Amazon Nova Canvas MCP Server

Overview

Amazon Nova Canvas MCP Server is a Model Context Protocol (MCP) server for generating images via Amazon Nova Canvas. It is designed to be used by conversational assistants and creative agents, providing text-based and color-guided image generation with flexible parameters and AWS-integrated authentication.

• Category: Media Processing MCP Servers
• Brand: Amazon Web Services
• Source: https://awslabs.github.io/mcp/servers/nova-canvas-mcp-server

Features

Text-based image generation

• Generate images from natural-language text prompts using the generate_image tool.
• Configure image dimensions in the range 320–4096 px.
• Choose image quality options (e.g., higher quality vs. faster/cheaper, as provided by Nova Canvas).
• Use negative prompting to specify what should be avoided in the output.
• Generate multiple images per request (1–5 images).
• Control guidance with cfg_scale parameter (range 1.1–10.0) to tune prompt adherence vs. creativity.
• Use seeded generation for reproducible outputs.

Color-guided image generation

• Generate images guided by color palettes using the generate_image_with_colors tool.
• Specify up to 10 hex color values to shape the image’s style, palette, and mood.
• Inherits the same customization options as text-based generation:
  • Dimensions (320–4096 px)
  • Quality options
  • Negative prompts
  • Multiple images per request (1–5)
  • cfg_scale control
  • Seeded generation

Workspace integration

• Save generated images into user-specified workspace directories.
• Automatic folder creation if the target directory does not exist.

AWS authentication & integration

• Uses AWS profiles (e.g., via AWS_PROFILE) for secure access.
• Connects to Amazon Bedrock and Amazon Nova Canvas services.
• Relies on standard AWS credential mechanisms (e.g., aws configure or environment variables).

Prerequisites

• uv installed (from Astral or GitHub).
• Python 3.10 installed via uv python install 3.10.
• AWS account with:
  • Amazon Bedrock enabled.
  • Amazon Nova Canvas enabled.
• AWS credentials configured with permissions for Amazon Bedrock and Nova Canvas (via aws configure or environment variables, with a suitable IAM role/user).

Installation & Configuration

General MCP configuration example (e.g., Amazon Q Developer CLI)

• Configure in your MCP client (such as ~/.aws/amazonq/mcp.json) to run via uvx, including environment variables:
  • AWS_PROFILE
  • AWS_REGION
  • FASTMCP_LOG_LEVEL
• Server is typically configured as a stdio MCP server with awslabs.nova-canvas-mcp-server@latest as the tool reference.

Windows installation

• Use uv with tool run --from awslabs.nova-canvas-mcp-server@latest awslabs.nova-canvas-mcp-server.exe.
• Configure MCP server entry with:
  • type: stdio
  • timeout (e.g., 60 seconds)
  • Environment variables: AWS_PROFILE, AWS_REGION, FASTMCP_LOG_LEVEL.

Docker

• Can be run via Docker after building the image:
  • docker build -t awslabs/nova-canvas-mcp-server .
  • (Then configure MCP client to talk to the container, as appropriate.)

Pricing

• The documentation provided does not specify pricing for the Amazon Nova Canvas MCP Server itself.
• Usage costs are likely governed by standard AWS pricing for Amazon Bedrock and Amazon Nova Canvas; consult AWS pricing pages for details.