Hey, fair questions

Copilot and Claude are good tools. What MergeMonkey does differently is how it gets to the result. The system marks signals and areas of concern first using static analysis, then the LLM verifies those, traces logic and data flow across the repo, and also does its own discovery on top of that. But because it starts with pre-collected signals instead of figuring out everything from scratch, the LLM's workload is way smaller. That's why Claude's code review costs an estimated $15-25 in tokens per PR. MergeMonkey does it in cents.

The other thing is flexibility. Most tools charge a fixed per-seat price and you don't really know what's happening behind the scenes. With MergeMonkey you pay a small fee for the tool but bring your own API key and model through OpenRouter. No black box. Tools like Copilot stay on affordable models even when better ones come out because that's how they keep margins. With MergeMonkey it's your choice, you pick whatever model you want.

There's a company using it right now as pilot, 20 devs, and their API cost for a month came out to about 11 euros for 20devs. CodeRabbit and Copilot charge $20-25 per seat. The hybrid approach gives the same level of results without burning through tokens because the LLM isn't doing all the heavy lifting.

Here's a real review on my professor's open source repo: NMFCode/NMF#134

I'm still improving it every day and looking for feedback from more people. If you want to give it a try, I'd appreciate it. If you like it, give me feedback so I can keep making it better. If you don't, removing it takes one click.

Happy to reply any question

If you want, just give it a 7 day try first? then its on you guys later want to keep or remove it

Sorry, I don't have time for this now.
I hoped having an example of PR review made by that tool would give me understanding of how quality the review is, but it hasn't given me such understanding..

@YuriiMotov
Your criticism was on point so I tried collecting some proof and asked Martian for a bench. Luckily they did, and based on the results I'll be placed 4th, beating Claude, Copilot and CodeRabbit. Unfortunately they're asking for some online presence first - till now I was only focusing on private pilots. This is proof of my claim:
https://discord.com/channels/1449120660877086824/1488625292518887464/1490291820708761765

If you're still not ready to give it a chance, sorry for wasting your time. Trying to pursue you more will make me look like I'm spamming you which would feel worse for me at this point.

I don't think I will try it, sorry..

I just accidentally found this PR review Mergemonkey-demo#5, and unfortunately I don't find it useful for me..
It looks more like an overview than a real review. Also, bad formatting makes it hard to read..
I prefer Copilot or Claude Code with my custom prompt emphasizing the things that I expect it to pay special attention to.

As for price, Copilot or Claude Code are free for me. At least for now.

Thanks, Yuri. I won't bother you anymore, but I'll continue making these kinds of testing in the fork. So if you ever want to take a look, feel free to do so - your feedback is really helpful to me.
Mergemonkey-demo#1