Lets agents generate images from text prompts and edit existing images with prompts and masks using OpenAI gpt-image-1, saving results to disk.
接入配置
选择客户端，复制配置，再按项目文档补齐环境变量。
claude mcp add gpt-image-1 --env GPT_IMAGE_OUTPUT_DIR='<GPT_IMAGE_OUTPUT_DIR>' --env OPENAI_API_KEY='<OPENAI_API_KEY>' -- npx -y @cloudwerxlab/gpt-image-1-mcp~/.codex/config.toml
[mcp_servers.gpt-image-1]
command = "npx"
args = ["-y", "@cloudwerxlab/gpt-image-1-mcp"]
env = { GPT_IMAGE_OUTPUT_DIR = "<GPT_IMAGE_OUTPUT_DIR>", OPENAI_API_KEY = "<OPENAI_API_KEY>" }
# codex mcp add gpt-image-1 --env GPT_IMAGE_OUTPUT_DIR='<GPT_IMAGE_OUTPUT_DIR>' --env OPENAI_API_KEY='<OPENAI_API_KEY>' -- npx -y @cloudwerxlab/gpt-image-1-mcp~/.cursor/mcp.json
{
"mcpServers": {
"gpt-image-1": {
"command": "npx",
"args": [
"-y",
"@cloudwerxlab/gpt-image-1-mcp"
],
"env": {
"GPT_IMAGE_OUTPUT_DIR": "<GPT_IMAGE_OUTPUT_DIR>",
"OPENAI_API_KEY": "<OPENAI_API_KEY>"
}
}
}
}.vscode/mcp.json
{
"servers": {
"gpt-image-1": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@cloudwerxlab/gpt-image-1-mcp"
],
"env": {
"GPT_IMAGE_OUTPUT_DIR": "<GPT_IMAGE_OUTPUT_DIR>",
"OPENAI_API_KEY": "<OPENAI_API_KEY>"
}
}
}
}~/.gemini/settings.json
{
"mcpServers": {
"gpt-image-1": {
"command": "npx",
"args": [
"-y",
"@cloudwerxlab/gpt-image-1-mcp"
],
"env": {
"GPT_IMAGE_OUTPUT_DIR": "<GPT_IMAGE_OUTPUT_DIR>",
"OPENAI_API_KEY": "<OPENAI_API_KEY>"
}
}
}
}claude_desktop_config.json
{
"mcpServers": {
"gpt-image-1": {
"command": "npx",
"args": [
"-y",
"@cloudwerxlab/gpt-image-1-mcp"
],
"env": {
"GPT_IMAGE_OUTPUT_DIR": "<GPT_IMAGE_OUTPUT_DIR>",
"OPENAI_API_KEY": "<OPENAI_API_KEY>"
}
}
}
}.trae/mcp.json
{
"mcpServers": {
"gpt-image-1": {
"command": "npx",
"args": [
"-y",
"@cloudwerxlab/gpt-image-1-mcp"
],
"env": {
"GPT_IMAGE_OUTPUT_DIR": "<GPT_IMAGE_OUTPUT_DIR>",
"OPENAI_API_KEY": "<OPENAI_API_KEY>"
}
}
}
}需要的环境变量
GPT_IMAGE_OUTPUT_DIROPENAI_API_KEY
把占位值换成你自己的密钥；不要把密钥提交到代码仓库。