mcp tool calls broken due to enforcing a strict 60s timeout #7324
Description
What happened?
something changed between the 0.1.22 and 0.2.1 releases, gemini-cli seems to be enforcing a 60s timeout on mcp tool calls. essentially the timeout field in mcp server config is being ignored. this seems at a very low level in mcp that can't be worked around in the tool implementation.
use case: i created an internal mcp server for our build/test integration which takes 2-3 minutes to complete for each call.
this enforcement is completely breaking the workflow, since gemini can't get the tool output for context and continue with its fixing loop.
What did you expect to happen?
mcp client should wait for the tool call to complete, and honor the mcp server timeout field. the mcp server sends notifications/progress periodically, which should provide enough context for the mcp client to keep waiting for the server output.
Client information
Client Information
Run gemini to enter the interactive CLI, then run the /about command.
> /about
│ About Gemini CLI │
│ │
│ CLI Version 0.2.2 │
│ Git Commit 32204dca │
│ Model gemini-2.5-pro │
│ Sandbox no sandbox │
│ OS win32 │
│ Auth Method gemini-api-key │
│ GCP Project <redacted> │Login information
No response
Anything else we need to know?
No response