grep tool times out on large repos before returning any results

[mcodilla]

Describe the bug

When using the built in grep tool in a large mono repo, the tool returns "timeout" with no results. I had to instruct copilot to use ripgrep rg, instead.

Affected version

GitHub Copilot CLI 1.0.36

Steps to reproduce the behavior

1. Open Copilot CLI in a large monorepo
2. Ask to search for a pattern across a folder and its subdirectories, dont nudge it to use ripgrep yet
3. It'll start grepping and maybe timeout

Expected behavior

The grep tool should return results (even partial) within a reasonable time

Additional context

No response

[rikogeln-havok]

We are running into this issue on large codebases. e.g 500k+ files in large repository 30% of grep calls time out, agent (opus 4.6) might even retry after and timeout again - 2 x 20s timeout with no result at all.

I see multiple mitigations one could take:

• good .gitignore file - pitfall with perforce p4: need to translate p4ignore into a gitignore or .ignore/.rignore file
• add more irrelevant directories to .ignore/.rgignore 
• use an extension to pre-tool hook global grep and slap the agent when it does unscoped greps

unscope grep slap extension might be the biggest fastest win on very big repos.
a message that it timed out and controllable timeout could also help.

github cli team - contact me for OTEL sessions showing the issue