Gemini Mcp
Generate and edit marketing or UI images through Google Gemini image models directly from your coding agent’s toolchain.
Overview
Gemini MCP is an MCP server for the Build phase that generates and edits images using Google Gemini Nano Banana image models from your agent.
What is this MCP server?
- Generate images with Google Gemini Nano Banana and Nano Banana Pro image models
- Edit existing images with configurable GEMINI_IMAGE_MODEL override (default gemini-3-pro-image)
- Required GEMINI_API_KEY from Google AI Studio
- GEMINI_OUTPUT_DIR and GEMINI_INPUT_DIR control save locations and input filename resolution
- Stdio @chrischall/gemini-mcp v0.4.0 on npm
- Server version 0.4.0
- Default image model gemini-3-pro-image (overridable via GEMINI_IMAGE_MODEL)
- 4 environment variables (1 required secret GEMINI_API_KEY)
What problem does it solve?
You need landing and product visuals while coding but switching to a separate image UI breaks flow and makes versioned asset paths hard to manage.
Who is it for?
Indie builders who already use Google AI Studio keys and want agent-driven image iterations for sites, apps, and launch graphics.
Skip if: Teams requiring strict brand templates, non-Google image stacks, or production pipelines without API cost controls.
What do I get? / Deliverables
After registration with a Gemini API key, your agent can generate and edit images into directories you configure alongside your codebase.
- Generated image files on disk from agent prompts
- Edited images using Gemini image models with optional input directory resolution
- Configurable model and output paths for repeatable asset workflows
Recommended MCP Servers
Journey fit
Visual assets for landing pages, app store shots, and demos are produced during Build when you integrate creative tools into the product workflow. Integrations subphase matches wiring an external Gemini image API into MCP so agents can call generation and edit operations programmatically.
How it compares
Gemini image API MCP wrapper, not a Figma plugin or standalone design agent skill.
Common Questions / FAQ
Who is gemini-mcp for?
Solo developers and content-heavy product builders who want Claude or Cursor to call Google Gemini image models for generate-and-edit tasks.
When should I use gemini-mcp?
Use it during Build when you need quick marketing art, UI mock imagery, or edited assets saved to project folders via MCP.
How do I add gemini-mcp to my agent?
Install @chrischall/gemini-mcp, add it as a stdio MCP server, set GEMINI_API_KEY and optional GEMINI_OUTPUT_DIR or GEMINI_IMAGE_MODEL, then invoke image tools from your agent.