codex-openai-image-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| OPENAI_API_KEY | No | Required when IMAGEGEN_PROVIDER is 'openai-api'. API keys are read from the environment only. | |
| IMAGEGEN_OUT_DIR | No | Output directory for generated images, relative to the working folder. | generated-images |
| IMAGEGEN_PROVIDER | No | Selects the image backend. Defaults to 'mock'; set 'codex-cli' for Codex CLI or 'openai-api' for the OpenAI Images API. | mock |
| IMAGEGEN_CODEX_BIN | No | Path or command name for the Codex CLI binary. | codex |
| IMAGEGEN_READ_ROOT | No | Root folder that input images must be inside. Defaults to the working folder. | |
| IMAGEGEN_TIMEOUT_SEC | No | Timeout in seconds per requested image. | 300 |
| IMAGEGEN_MAX_PER_CALL | No | Maximum number of images one call can generate. | 8 |
| IMAGEGEN_OPENAI_MODEL | No | OpenAI image model to use for the openai-api backend. | gpt-image-2.5-sunburst |
| IMAGEGEN_PREVIEW_SIZE | No | Preview size in pixels for returned image previews. | 512 |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| generate_imageA | Generate one or more images from a text prompt. Saves full-size PNGs (plus a .json sidecar with the prompt and settings) in the output folder and returns their paths with small previews. |
| edit_imageA | Change an existing image with a text instruction (e.g. 'make the sky a sunset', 'remove the person on the left', 'turn it into a watercolour', 'remove the background'). The original is never modified; edited copies are saved in the output folder with a .json sidecar, and their paths are returned with small previews. |
| generate_setA | Generate several different images that share one style, e.g. a set of icons, stickers, badges, character portraits or story illustrations. Each item gets its own file, named after the item, in one folder. Returns paths and previews. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 3 tools
Each tool targets a distinct operation: editing an existing image, generating a coordinated set, and generating standalone images. The only mild overlap is that generate_image says 'one or more images' while generate_set also produces multiple images, but the shared-style/folder semantics distinguish them reasonably well.
All three tools follow a consistent verb_noun snake_case pattern: edit_image, generate_set, generate_image. The convention is predictable and readable.
Three tools is a focused, well-scoped surface for an image generation/editing server. It is slightly thin but each tool clearly earns its place without redundancy.
The core lifecycle is covered: generate single images, generate coordinated sets, and edit existing images. Minor gaps exist around image variations, upscaling, or format conversion, but agents can work around them for the stated purpose.