Last Notes
Dev/Hack/Day + Cypherpunk Meetup =
@nprofile…2za5
If you're coming to @nprofile…8er3, don't miss it.
Get your early bird tickets at
freedomtechsummit.com
do you want an age verification?
I'm probably wrong but I interpret Alex's post as a kind, ironical, and provocative invitation for improvement of DMs implementation.
At least to me, something doesn't feel right about White Noise yet, but it's close to what Signal (and SimpleX) do in terms of privacy.
#nevent1q…casu
there is also the absence of support for the necessary primitives in nip-07 signers for marmot MLS. i've specced something up and implemented it. a full browser based implementation is coming thoon, including that signer. and a full Go implementation. the javascript implementation is through a "tinyjs" build target i'm mostly done building that lets you do a whole web app in tinygo.
@nprofile…6lrd If the related post is still available please share if you can, thanks. I'm curious what didn't work for you.
https://image.nostr.build/f6683c508540ef5048a8c5270f8db3b0e48a88e4be8763067ab7af774389c2f2.png
https://video.nostr.build/f4338b05c0b69836410afd9042b889885ffbc6892b10802efad3ebc6c901000f.mp4
you know this for this many days and you are still using cake? damn why tho
Especially if you are dyslexic, which book narration software do you prefer?
I'd like something FOSS for GNU/Linux that can ideally do well
- reading with expression
- speedup up to ~4x
- properly parse weird doc structure with footers/headers/columns/etc.
- (in a non-disruptive way) notify when a related picture becomes visible
- the more supported formats, the better
Easier to vibe code it already? Nothing worked for me in the past.
I keep reading manually when I have capacity. Still a torture to me.
#asknostr
you are welcome. 🫂 Humans are "discomfort avoidance". We avoid it if we can. Sometimes it works and sometimes it makes it worse. There is no one handbook but only trial and error. At the end of the day, how you want to earn respect defines how you respect yourself. 🙏
That's a lot of fruiting blocks!
They build apartheid communities: bribing officials and excluding or treating locals as second class.
This is why @nprofile…mvq5 exists
No gatekeepers
Sats are access
It’s only 'dubious' if you actually trust the marketing. I treat Telegram like a public square: lots of noise, zero trust. That's why my signal here. Telegram is for the network effect.
I got your question loud and in clear-text, by the way.
We are the best firewall 💯
Idk. Still have to work 😆
Someone impersonating @npub10vl…sp42 tried to phish me last night over Telegram but failed to do so.
Don’t trust, verify. No exceptions.
#BeTheHumanFirewall
Good morning Nostr ☕️
I’ve successfully added support for about 3 Nostr NIPs to Claude Code.
I took care of redacting free form fields (content, tags) to prevent prompt injection from pubkeys not manually allowed.
#coffeechain
Check if you were impacted:
https://securityaffairs.com/189948/hacking/malicious-litellm-versions-linked-to-teampcp-supply-chain-attack.html
To answer your question directly: I am building my own “OpenClaw” with Claude instead of adopting thousands of lines of code I can’t possibly review or understand.
This video explains how to use it, in case you were interested:
https://youtu.be/qKU-e0x2EmE
I had the same problem. Staying humble was easy. Stacking sats was hard.
Has anyone tried Andrej Karpathy’s Autoresearch?
I’m creating a Nostr plugin using /create-agent-skill skills authored by Taché and inspiring it on Claude’s official Telegram plugin. This Claude Code plugin uses the new Channels feature that Claude Code has.
github.com/glittercowboy/taches-cc-resources
you ask too much
🚨🚨🚨🚔🚔🚔👮♂️👮♂️👮♂️
> they made a blog something about how they couldn't set protected flags because of lack of support in relays
Interesting. Could you possibly share a link please? I couldn't find the post; hopefully it's not deleted.
give me one, ohh so dont have one either, now we are even then lol
ahh now you are too high for snitching, you are lame
okay only if you tell me more about what i am doing on my account then you will get some 69 jokes on your peepee
ahh i see thank you for your help sir you are really doing a great job for us
he warned y'all guys, dont click on this douchebags links
https://blossom.primal.net/18722f37ab0ed22da823dc78c2c5b1d7fe45624d01d9d1e784aba769af237acf.jpg
> "we kill based on metadata" literal quote from the NSA or CIA i forget which
Michael Hayden, former director of both NSA and CIA. "We kill people based on metadata." Said it publicly, on the record, as a point of pride about the drone program.
Content encryption is locking your diary while wearing a GPS ankle monitor. WhiteNoise encrypts the diary and broadcasts the ankle monitor data to every relay on the network.
> nuff said
https://npub1fjqqy4a93z5zsjwsfxqhc2764kvykfdyttvldkkkdera8dr78vhsmmleku.blossom.band/34818d8c85791156e85e7915aab3920f1d92a696d347a85e55cd0304449b15a8.png
They should collet it in bitcoin but the heat would be too much rn.
I think they're going for the opposite. They are building it so that all events are wide open access. The relays will not be able to ask for AUTH, nor do any kind of rate limits to protect from DoS (other than IP limits, which are flawed and hurt the ToR).
AUTH could be added in to the client, but it likely won't. BUT if it did, then you could have say, per-device auth keys, or keys not tied to any nostr id, just that registration will be harder that way and the relay still cannot protect because it does not know which groups are what. The addition of this complication makes it sound nearly impossible to add later.
So yeah, using auth with it. Seems like a non-starter to me.. Designed like this on purpose. A huge message broker free-for-all to read and write all groups.
authmageddon comes, my relay dev/operator frens.
claude:
You're not misreading it. MLS encrypts content, not metadata. The relay is the delivery infrastructure — it has to know enough to route. The question is never "does the relay know things" but "who else does."
Open relay: the relay knows AND the entire internet knows. Auth relay: the relay knows, period. WhiteNoise's approach: the relay knows AND the internet knows, but they rotate keys so the content stays encrypted. The metadata — who's posting, when, how often, group size, membership changes via welcome messages and key packages — is naked either way on an open relay.
So the real trust decision is: do you trust one relay operator, or do you trust the entire internet to not do traffic analysis? That's not a hard choice. Auth relay with a known operator is strictly better than open relay with rotating keys. The only argument against is the ideological one Cloudfodder already identified: "we don't want the relay to know." But the relay already knows. You're just choosing whether everyone else also knows.
The bridge model makes this explicit. The bridge has one relay in its list. That relay requires auth. Trust is declared, bounded, and auditable. WhiteNoise's model pretends trust isn't required and ends up trusting everyone.