What it does
The server exposes four tools. generate_image creates images from a text prompt with DALL-E 2 or DALL-E 3, with selectable sizes, up to 10 images per call, and quality and style options for DALL-E 3. edit_image changes an existing image from a prompt, with an optional mask, and create_variation produces variations of an existing image; both work with DALL-E 2 only. validate_key checks the OpenAI API key.
The README describes use with Cline in VS Code. For Cline, it recommends setting saveDir to the current workspace directory so the generated images can be located and displayed in the conversation. Outside Cline, the same tools can be used from any MCP-compatible client that can run the server.
Setup involves building the server and adding it to the MCP settings JSON, pointing to the built index.js file and supplying an OpenAI API key. The output directory can be set per call or through SAVE_DIR in a .env file. Development and testing need a .env file in the project root containing the key. The test suite makes real API calls, and a script generates a sample image in the assets directory for the edit and variation tests.
Limitations and requirements
Requires an OpenAI API key. The edit_image tool is subject to a known issue, reported on 18 March 2025, in which the DALL-E 2 edit API sometimes ignores the prompt and returns the original image; the README suggests create_variation as an alternative. Editing and variations are supported for DALL-E 2 only. Tests use real API calls that may incur charges on the OpenAI account.
Safety
Our team has not reviewed this resource. Only automated rules checked it; resources with high-risk signals are not published. These are signals, not a guarantee that installing it is safe. Try it in a sandbox project first.
- Runs code or changes an assistant's behaviour
Sources
- Cline MCP Marketplace first seen 13 Nov 2026 · last seen 18 Nov 2026 · its page on the source
garoth/dalle-mcp
As shown on cline.bot when we last read it