Known issues

This document provides details and workarounds for known issues with Google and Google Cloud remote Model Context Protocol (MCP) servers.

Authentication and Authorization Issues

The following known issues involve problems with MCP authentication and authorization.

Antigravity doesn't support MCP OAuth specifications

Antigravity doesn't support the MCP OAuth specifications, which prevents you from using a client ID and secret. You can configure Google and Google Cloud remote MCP servers in Antigravity and see the tools. However, Antigravity doesn't provide a way to use a client ID and secret or agent identity, so you can't authenticate to the MCP server.

Tool issues

The following known issues affect specific Google or Google Cloud MCP tools.

Compute Engine `list_images` tool returns permission denied

The Compute Engine list_images tool fails with a permission error when querying public images like fedora-cloud.

Client-specific issues

The following issues involve problems with specific MCP clients.

Gemini CLI gives API error

The following issue occurs when the name of the MCP server is longer than 64 characters or contains characters other than alphanumeric characters, underscores, dots, colons, or dashes:

[API Error: {"error":{"message":"{\"error\": {\"code\": 400, \"status\": 400, \"message\": \"* GenerateContentRequest.tools[0].function_declarations[14].name: Invalid
function name. Must start with a letter or an underscore. Must be alphanumeric (a-z, A-Z, 0-9), underscores (_), dots (.), colons (:), or dashes (-), with a maximum length
  of 64.\\n*

To resolve this issue, change the name of the MCP server in your Gemini extension file to comply with the required naming conventions. Gemini CLI extension files are typically located in: ~/.gemini/extensions/EXT_NAME/gemini-extension.json where ~/ is your home directory and EXT_NAME is the name you gave the extension. For more information, see Configure MCP in an AI application