Skip to content
Discussion options

You must be logged in to vote

Hi — thanks for reporting this!

The issue occurred because older versions of Rapid-MLX only recognized the legacy servers key, causing standard mcpServers configurations to silently load as 0 servers.

This is fixed in the latest release (v0.12.17), which supports both formats and now surfaces configuration errors properly.

Please upgrade:

pip install -U rapid-mlx  # or brew upgrade rapid-mlx

Ensure your mcp.json is structured like this:

{
  "mcpServers": {
    "filesystem": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/allowed/directory"]
    }
  }
}

Then start your server and verify the connection:

rapid-mlx serve qwen3.5-4b --mcp-c…

Replies: 1 comment

Comment options

You must be logged in to vote
0 replies
Answer selected by raullenchai
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Category
Q&A
Labels
None yet
2 participants