Images

You run an MCP server dedicated to AI-powered image generation, capable of using Stability AI and Black Forest Labs models. It accepts prompts from an MCP client, generates images, and keeps detailed provenance and flexible filenames for easy organization and auditing.

How to use

You integrate this server with any MCP-compatible client to generate images from text prompts. You can choose between Stability AI models (e.g., SD series) and Black Forest Labs Flux models, controlling aspects like model, aspect ratio, and generation parameters. When you call the image-generation tool, you provide a prompt and optional parameters such as a negative prompt, a specific model, and the output filename template. The server returns the generated image along with metadata that records the generation settings, provenance, and checksums for integrity.

Practical usage tips:

• Choose the provider you want to use: stability or bfl. The default is stability if you omit the provider.

• Use a descriptive prompt and, if needed, a negative_prompt to steer the result away from undesired features (supported by compatible providers).

• Set a filename_template to organize outputs. You can embed timestamp, provider, model, and a subject derived from the prompt to keep outputs searchable.

How to install

# Clone the images MCP server repository
git clone https://github.com/rmrfslashbin/mcp-server-images.git
cd mcp-server-images

# Install and synchronize dependencies
uv sync

Set up the required API keys in your environment before starting the server.

# Example environment variables
export STABILITY_API_KEY=sk-...
export BFL_API_KEY=...

Run the server using the standard MCP start command.

uv run mcp-server-images
# or, if you prefer Python module execution
python -m mcp_server_images

Additional configuration and notes

Configuration is primarily via environment variables and filename templates. The server supports multiple providers and models, with a flexible template system for output filenames. You can tailor prompts, seeds, and CFG-scale where supported by the selected provider.

Security and credentials: keep your API keys secure and do not expose them in client requests. Use restricted access to the MCP endpoint and rotate keys if you suspect exposure.

Troubleshooting and tips

If image generation fails, check that API keys are valid and that the selected provider/model is supported. Review the generated metadata to understand parameter values used during generation and to verify provenance.

Ensure your output directory exists or is writable by the server, and verify that your filename_template produces valid filenames in your environment.

Notes on usage with MCP clients

This server works with any MCP-compatible client. You can integrate it with chat interfaces, editor extensions, or other MCP-enabled tools to generate images directly from prompts without building a custom bridge.

Available tools

generate_image

Generate images from text prompts with AI-optimized prompting. Accepts optional parameters like negative_prompt, provider, model, aspect_ratio, cfg_scale, seed, output_dir, and filename_template.