Initial Checks

• I confirm that I'm using the latest version of Pydantic AI
• I confirm that I searched for my issue in https://github.com/pydantic/pydantic-ai/issues before opening this issue

Description

I'm using the following code to create an MCP server:

file_system_mcp = MCPServerStdio(
    'docker',
    args=[
        'run',
        '-i',
        '--rm',
        "--mount",
        "type=bind,src=...",
        "-w", "/src",
        'mcp/filesystem',
        "...",
    ],
)

However I forgot to start Docker deamon.

Later when I use async with file_system_mcp: I get the exception:

unhandled errors in a TaskGroup (1 sub-exception)

instead of a helpful exception such as:

docker: Cannot connect to the Docker daemon at unix:///Users/jmonsonego.ctr/.docker/run/docker.sock. Is the docker daemon running?

I can see the mentioned sub exception as McpError('Connection closed') when I view the local variables during debugging, but that's not specific enough.

Why doesn't Pydantic propagate the exceptions that happen during the __aenter__ of the MCP context manager outside?

Example Code

Python, Pydantic AI & LLM client version

Python 3.13
Pydantic AI 0.7.3