Copilot SDK doesn't recognize the Environment variables from MCP server's configuration

[SangTruongTan]

I'm developing a web app that use Copilot SDK as a backend AI engine. However, while configuring the MCP server. I'm facing an issue that the environment variables that aren't recognized and passed into the MCP process. It's weird because my MCP server is functioning properly with the native Copilot CLI. I traced the source code and found that Copilot SDK is just passing all the configuration to the Copilot CLI instance through JsonRPC. So, I think there should be either an interface mismatch or Copilot CLI issue.

    mcp_servers: dict[str, MCPLocalServerConfig] = {
        "gerrit-mcp": {
            "type": "local",
            "command": "python",
            "tools": ["*"],
            "args": [
                "<path_to_mcp>/mcp-gerrit/src/server.py",
                "--log-file",
                "./gerrit_mcp.log",
                "--log-level",
                "DEBUG",
            ],
            "env": {
                "GERRIT_URL": "<url>",
                "GERRIT_USERNAME": "<user>",
                "GERRIT_PASSWORD": "<password>",
            }
        }
    }

    client = CopilotClient()
    await client.start()

    session = await client.create_session(
        {"model": "gpt-4.1", "streaming": True, "mcp_servers": mcp_servers}
    )

[SangTruongTan]

Can anybody explain why this issue is closed? If it's duplicated, please provide the issue id. If it's completed, please explain why it's not working.