@bytebot
Today I have two thoughts on Mitsukeru. Well ok, three. The last thought might be the most interesting.
1. This is the week of 5th January - so there will be some beta testing for investors as things get packaged. App Store signing takes too long since they have a backlog and also I think a lot of holiday vibecoded apps to go thru. So this week, investors get access to the Chrome plugin to sideload, and also, maybe just a compiled, unsigned macOS app. It is a beta, after all.
As mentioned earlier, before 31 Jan 2026 is TGE, because I personally want to be at FOSDEM!
2. Comprehensive protection against everything that runs on your macOS or Windows box is actually a very complex topic. ICP today is the crypto user, so the Clipboard Address Guard is timely, and useful, for said user. As is LinkSentinel.
Mitsukeru is a collection of tools, like Microsoft PowerToys. Pick and mix, and eventually you'll get one big fat binary, like CleanMyMac.
I personally use browser agents *carefully* - be it Comet, or with Claude and its ability to control Chrome. Then there are also Playwright MCPs. All this can be memory and CPU hungry, in addition to watch over you, cycle wise.
I just casted about Notion AI being a potential loophole. So it is imperative to look at firewalls, a lot stricter. With a Linux hat, this is SELinux (contexts) and ipchains/iptables/firewalld.
One this is certain - we can't monitor all software. We can only monitor dark patterns. And this is how we catch malware.
3. The Reputation Layer. I have casted often about this. I have got some thoughtful DMs and conversations going, but no one is quite ready to say "let's take on Ethos, meld a bit of Wikipedia/Grokipedia/Internet Archive into it, and be a record". Not yet, anyway? Screenshots are super important, as a record. Previous handles. Etc. As usual I have some anecdotal data.
- there was some onchain analysis about RichardHeartWin and a pretty interesting thread... about him. He has a 1147 Ethos score, questionable, but it doesn't seem to affect him moving millios of funds, and having fresh targets. But what ethos truly lacks for his profile? Context. It is a lot of opinion, it doesn't have the rigour of references, like a Wikipedia article has.
- frankdegods is doing a comeback tour as an AI vibe coder, he is being dunked on a lot, has a 765 Ethos score, so clearly untrusted, but again if someone said "show us the receipts," it is completely lacking
- if we are talking about people "rebranding", we have NFT_GOD -> Alex Finn (also a vibecoding guru), and also Ryan Carson, builder in residence a AmpCode, who previously had an NFT fund, and also was the former COO of Proof and Moonbirds. Check them both out on Ethos. Alex lack's his previous NFT days (its focused on him doing AI stuff), and Ryan has 1015/Questionable but he has long left the Web 3 space - but again, zero context about him.
- and today, we have henloitsjoyce, 1019 Ethos score, Questionable, looking for accommodation in SF, people bringing up her past, and at the same time, we can see kind of a receipt via a proper Ethos slash. I say kind of because they all go to X links. Which are not permanent records, like the Internet Archive. So people may or may not remember her fake AI agent (which was then acquired by another company), the SOL rugged, or what she's done since - but she is at Crossmint, and is still very much at large.
- but one more thing, even here, on Farcaster, we have farcsteradmin being called out over Merv. Enough noise was made, and the person making the noise got the refund so the casts are now deleted, which is fair. But there are others. Neutral (1216) Ethos score, which is accepted.
So I firmly believe the reputation layer is missing, and a problem to solve. Does it sound too China-like with a social credit score? Maybe.
Not everything happens on the social layer for reputation's either. Or at least not in public (so discord/telegram chats, are good too). Imagine going to a telegram group, being called a scammer, and your post deleted - ok - but where is the proof that you're a scammer? Maybe this is LinkedIn but curated by your peers ;)
Still think this is a *good* problem to solve. Selfishly I've said, this helps Mitsukeru get better at protecting users. Why hasn't it been solved yet?
P/S: 2026 is the year of portable KYC as well. Added layer there for ones that are OK to ID verify.