Practical overview
When it fits
Use it when an MCP assistant workflow needs to create or edit image files and return them to the assistant, rather than when you need a standalone image-editing interface.
Who it is for
- People using MCP-compatible assistants who want image generation or editing within their existing workflow.
- Developers configuring Codex, Cursor, Claude Code, or another MCP client to call image providers.
Documented uses
- Generate an image from a text request, with optional prompt enhancement and quality selection.
- Request a change to an existing image by supplying its absolute path and an editing instruction.
- Choose among Gemini, OpenAI, and BytePlus Seedream for image generation and editing.
Requirements & limits
Setup requires Node.js 22 or later, an MCP-compatible client, and an API key for one supported image provider. The repository is MIT licensed.
- Gemini is the default provider; provider selection changes both prompt enhancement and image generation.
- Gemini supports 1K, 2K, and 4K sizes; OpenAI supports 1K, 2K, and 4K with PNG or JPEG; Seedream supports 1K and 2K with PNG or JPEG.
- Google Search grounding is available only with Gemini.
- OpenAI may require organization verification before image generation.
- Higher quality presets can take longer and cost more; results vary by provider.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Set up MCP Image with a Gemini key
Configure an MCP client to send image requests through MCP Image and save returned files to a chosen directory.
Before you start
- Node.js 22 or later
- An MCP-compatible client, such as Codex, Cursor, or Claude Code
- A Gemini API key
Steps
Set the Gemini API key as
GEMINI_API_KEYin the MCP server environment.Configure the client to run the command below and set
IMAGE_OUTPUT_DIRto an absolute path.npx -y mcp-imageRestart the MCP client after changing its configuration.
Ask the assistant to generate an image; the file is saved in the configured directory and returned as an MCP resource.
Expected outcome
The assistant can request image generation through MCP Image, with the resulting file saved locally and returned to the client.
Guide limitations
- A provider API key is required.
- The documented Gemini setup uses the default
fastquality preset unless changed. - Image results vary by provider.
Request an edit to an existing image
Send an image-editing instruction and a source image path through an MCP client.
Before you start
- MCP Image configured in an MCP-compatible client
- An API key for Gemini, OpenAI, or BytePlus Seedream
- The absolute path to the source image
Steps
Provide the assistant with an editing instruction and the absolute path to the source image.
Optionally select a provider or quality preset for the request.
The generated file is saved to the configured output directory and returned to the assistant.
Expected outcome
The MCP client receives the edited image file from the provider workflow.
Guide limitations
- The README documents the request pattern but does not provide provider-specific editing guarantees.
- Available image sizes and output formats differ by provider.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 14 October 2026
- Original website title
- shinpr/mcp-image · Official homepage title · retained in its source language