MCP server for sandboxed Python execution — bubblewrap on Linux, Docker on macOS. PEP 723 inline deps, multi-version Python, ephemeral environments.
Setup
Choose your client, copy the config, then fill in any required environment variables.
claude mcp add python-exec-sandbox -- uvx mcp-python-exec-sandbox~/.codex/config.toml
[mcp_servers.python-exec-sandbox]
command = "uvx"
args = ["mcp-python-exec-sandbox"]
# codex mcp add python-exec-sandbox -- uvx mcp-python-exec-sandbox~/.cursor/mcp.json
{
"mcpServers": {
"python-exec-sandbox": {
"command": "uvx",
"args": [
"mcp-python-exec-sandbox"
]
}
}
}.vscode/mcp.json
{
"servers": {
"python-exec-sandbox": {
"type": "stdio",
"command": "uvx",
"args": [
"mcp-python-exec-sandbox"
]
}
}
}~/.gemini/settings.json
{
"mcpServers": {
"python-exec-sandbox": {
"command": "uvx",
"args": [
"mcp-python-exec-sandbox"
]
}
}
}claude_desktop_config.json
{
"mcpServers": {
"python-exec-sandbox": {
"command": "uvx",
"args": [
"mcp-python-exec-sandbox"
]
}
}
}.trae/mcp.json
{
"mcpServers": {
"python-exec-sandbox": {
"command": "uvx",
"args": [
"mcp-python-exec-sandbox"
]
}
}
}