Connects Claude to three image generation APIs: Google Gemini (including the Nano Banana models), OpenAI's gpt-image, and BFL FLUX. Exposes a single generate_image tool that takes prompts, model selection, resolution (1K/2K/4K), aspect ratio, and optional input images for editing workflows. Google models support a mode flag to return generated images with descriptions, plus thinking controls. Handles path sandboxing when you set NANO_BANANA_OUTPUT_DIR, validating that input and output paths stay within bounds. Returns saved file paths and metadata. Reach for this when you want to give Claude multi-provider image generation without building separate integrations for each API.
claude mcp add --transport stdio mimo-3-mcp-imagenate uvx mcp-imagenate