Python server that exposes OpenAI's built-in web_search tool to other MCP clients.
Setup
Choose your client, copy the config, then fill in any required environment variables.
claude mcp add openai-websearch --env OPENAI_API_KEY='<OPENAI_API_KEY>' -- uvx openai-websearch-mcp~/.codex/config.toml
[mcp_servers.openai-websearch]
command = "uvx"
args = ["openai-websearch-mcp"]
env = { OPENAI_API_KEY = "<OPENAI_API_KEY>" }
# codex mcp add openai-websearch --env OPENAI_API_KEY='<OPENAI_API_KEY>' -- uvx openai-websearch-mcp~/.cursor/mcp.json
{
"mcpServers": {
"openai-websearch": {
"command": "uvx",
"args": [
"openai-websearch-mcp"
],
"env": {
"OPENAI_API_KEY": "<OPENAI_API_KEY>"
}
}
}
}.vscode/mcp.json
{
"servers": {
"openai-websearch": {
"type": "stdio",
"command": "uvx",
"args": [
"openai-websearch-mcp"
],
"env": {
"OPENAI_API_KEY": "<OPENAI_API_KEY>"
}
}
}
}~/.gemini/settings.json
{
"mcpServers": {
"openai-websearch": {
"command": "uvx",
"args": [
"openai-websearch-mcp"
],
"env": {
"OPENAI_API_KEY": "<OPENAI_API_KEY>"
}
}
}
}claude_desktop_config.json
{
"mcpServers": {
"openai-websearch": {
"command": "uvx",
"args": [
"openai-websearch-mcp"
],
"env": {
"OPENAI_API_KEY": "<OPENAI_API_KEY>"
}
}
}
}.trae/mcp.json
{
"mcpServers": {
"openai-websearch": {
"command": "uvx",
"args": [
"openai-websearch-mcp"
],
"env": {
"OPENAI_API_KEY": "<OPENAI_API_KEY>"
}
}
}
}Environment variables
OPENAI_API_KEY
Replace the placeholders with your own credentials and keep them out of version control.