`cursor-agent mcp list` fails on all or nothing basis

Where does the bug appear (feature/product)?

Cursor CLI

Describe the Bug

if a single MCP server from .cursor/mcp.json is unreachable, the cursor-agent mcp list command exits with a failure, even if the other server(s) are definitely working/able to be connected to.

verified that the order of mcp servers in .cursor/mcp.json does not matter

Steps to Reproduce

.cursor/mcp.json set up with multiple MCP servers. e.g.:

{
  "mcpServers": {
    "My Server": {
      "url": "http://127.0.0.1:3845/mcp" # not running/unreachable
    },
    "working-mcp-server": { # definitely works, but not when the unreachable server is alongside
      "command": "npx",
      "args": ["..."],
    }
  }
}
$ cursor-agent mcp list
# Failed to list MCP servers: fetch failed

removing unreachable server from .cursor/mcp.json

{
  "mcpServers": {
    "working-mcp-server": { # no change
      "command": "npx",
      "args": ["..."],
    }
  }
}
$ cursor-agent mcp list
# working-mcp-server: ready

Expected Behavior

The checks for connected MCP servers should fail gracefully and on an individual basis. If one server is unreachable, the other connections/statuses should not be affected.

Operating System

MacOS

Current Cursor Version (Menu → About Cursor → Copy)

$ cursor-agent --version                                                                                                                                                                           
-- 2025.11.06-8fe8a63

Does this stop you from using Cursor

Sometimes - I can sometimes use Cursor

Hey, thanks for the report. This is a known CLI issue - there are several related topics with the same error.

Your case highlights a specific aspect: lack of graceful degradation when one server fails. The team is already working on MCP error handling in the CLI.

Workaround: temporarily remove unreachable servers from your config.

This topic was automatically closed 22 days after the last reply. New replies are no longer allowed.