This is fantastic. Claude doesn't make it easy to inspect what it's sending - which would actually be really useful for refining the project-specific prompts.
Love you like it!! Let me know any ideas to improve it... I was thining in the direction of a file system and protocol for the md files, or dynamic context building. But would love to hear what you think.
> The quickest way to receive an invitation is to talk to someone you recognize from the site. If you wrote a link that was posted, please reach out in chat, we'd love to have you join the community. Finally, if you can't find anyone you know in the invitation tree and didn't author something posted to the site, consider getting to know the community in the chat room.
Not very difficult if you have an online presence somewhere you can use as proof that you will act in good-faith. Having a hacker news account for instance can make joining as easy as sending someone an e-mail (as I did) and asking. It is a website much more built upon trust, and so if you invite someone and they get banned for something, you are directly connected to that.
I like that you are treating peephole as a strict fallback after IR optimizations, with a tiny window and single pass. In a lot of compilers this stage turns into a junk drawer of pattern matches that quietly grow until no one remembers why half of them exist.
The opt_trace! hook is the underrated bit here. Once you start rewriting instruction sequences, knowing which patterns fire and how often on real programs is usually more valuable than another synthetic benchmark. Keeping it behind -O1 is a nice way to make sure you only pay for that complexity when users actually opt in.
