Question on PR review bots

[yashwant86]

Hi maintainers,

I noticed this project currently has a large number of open PRs, and I wanted to ask a general question about how you manage review load at that scale.

When review queues grow, do automated reviewers help reduce effort, or do they sometimes add more noise than signal?

I’m exploring a very lightweight review approach that does one thing by default:
it posts a short, structured summary of changes and stays silent otherwise.
Deeper analysis only happens if a maintainer explicitly asks for it.

If it’s useful, this can be tried on a single PR only, using free or lightweight models.
No workflow changes, and it’s easy to remove if it doesn’t help.

Completely understand if this isn’t a good fit, mostly trying to learn what actually helps when PR volume is high.

Thanks for all the work you put into maintaining the project.

[niheaven]

Hmm, that's a good idea imo.

[yashwant86]

Thanks for the feedback, glad it resonates.

If you’re open to it, I can share a link and explain exactly what would happen on a single PR. No obligation at all.

[niheaven]

Please share the link, and thanks for the help.

[yashwant86]

Thanks for the openness.

Here’s the link: https://mergemonkey.dev/

In short, what would happen:

• On a new commit, it posts one short, structured summary comment, optionally including a small Mermaid diagram.
• Nothing else runs unless someone explicitly asks for a deeper review using a PR comment command.

Happy to try this on just one PR first and remove it immediately if it’s not useful or gets in the way.

Appreciate you taking the time to consider it. Any feedback would be very helpful.