Qwen Image MCP server
{
"mcpServers": {
"acedatacloud-qwen-image": {
"command": "uvx",
"args": [
"mcp-qwen-image"
],
"env": {
"ACEDATACLOUD_API_TOKEN": "<acedatacloud_api_token>"
}
}
}
}{
"mcpServers": {
"acedatacloud-qwen-image": {
"command": "uvx",
"args": [
"mcp-qwen-image"
],
"env": {
"ACEDATACLOUD_API_TOKEN": "<acedatacloud_api_token>"
}
}
}
}{
"mcpServers": {
"acedatacloud-qwen-image": {
"command": "uvx",
"args": [
"mcp-qwen-image"
],
"env": {
"ACEDATACLOUD_API_TOKEN": "<acedatacloud_api_token>"
}
}
}
}{
"servers": {
"acedatacloud-qwen-image": {
"command": "uvx",
"args": [
"mcp-qwen-image"
],
"env": {
"ACEDATACLOUD_API_TOKEN": "${input:acedatacloud_api_token}"
}
}
},
"inputs": [
{
"id": "acedatacloud_api_token",
"type": "promptString",
"description": "Acedatacloud api token",
"password": true
}
]
}{
"mcpServers": {
"acedatacloud-qwen-image": {
"command": "uvx",
"args": [
"mcp-qwen-image"
],
"env": {
"ACEDATACLOUD_API_TOKEN": "<acedatacloud_api_token>"
}
}
}
}file .mcp.json (in your project root)file claude_desktop_config.jsonfile ~/.cursor/mcp.jsonfile .vscode/mcp.jsonfile ~/.codeium/windsurf/mcp_config.jsoninstall uvx mcp-qwen-imageversion 2026.8.24.1
What it does
MCP server for generating and editing images using Qwen Image 3 through Ace Data Cloud. It allows developers to integrate Qwen's image generation and editing capabilities into Claude or other AI applications.
- Generate 1 to 6 images from text prompts using Qwen Image 3 models.
- Edit images using 1 to 3 reference images as guidance.
- Query the status of asynchronous image generation or editing tasks.
Tools this server exposes
| Tool | Does |
|---|---|
| qwen_image_generate | Generate 1–6 images from text prompts. |
| qwen_image_edit | Edit images using 1–3 reference images. |
| qwen_image_get_task | Query the status of a single asynchronous task. |
| qwen_image_get_tasks_batch | Query the status of multiple asynchronous tasks. |
| qwen_image_list_models | List available models and compare Standard and Pro variants. |
Before you enable it
An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.