You'd think after setting up multiple nodes I'd quit being so impatient...sync status: 88%

StartOS machine up and it is so seamless...blockchain syncing...

Eagerly awaiting for my 2tb nvme to arrive so i can give StartOS 0.4 a try...

Been working on a daemon that monitors other services, autoheals issues, auto fixes and optimized code issues, and learns system behavior to prevent future issues. been running on my local server for a bit over a week so time to start testing in on production vps servers. If all goes well, will release soon. No external apis required.

Why have I been so quiet lately? I've been working on an AI agent framework to run my day-to-day. It keeps me up to date on news and Nostr. It handles all my e-mail. It handles all my finances. So what do I do with my days now? I spend time with family. I read books. I code. I travel. I build my citadel. Will I ever release it? Maybe. Eventually.

yo @jack mallers, more "What Grinds My Gears" please...maybe even a full episode of just that. Glad at least one person keeps it real on the pod circuit.

Been pretty quiet this week on Nostr as I've been building a trading bot. I'll post some more on this soon, it has been a fascinating undertaking. Honestly, it's been nice not being on any sort of social so I'll probably just be poking my head in here every once in a while moving forward. I guess the best place to follow now will be on Github lol. With that said, new Signet release in a week or two where I'll finally release Android signer support which has been working really great, even if I am a bit disillusioned by the failure of so many devs to follow standards and refusing PRs.

Finney is ready for testing. It is a simple Bitcoin dashboard for those of you that use Bitcoin as money. Currently, both Claude and FRED API keys are required. Claude is used to generate daily market insights which are also configurable to run on on % price change or manually. Digests cost about $0.01/digest.

GitHub

GitHub - Letdown2491/finney

Contribute to Letdown2491/finney development by creating an account on GitHub.

Pushed an update to trustedrelays.xyz to improve memory handling...if you want to run your own instance, i recommend 2gb RAM.

GitHub

GitHub - Letdown2491/trustedrelays: Reference implementation for Trusted Relay Assertions, trust scores for Nostr relays

Reference implementation for Trusted Relay Assertions, trust scores for Nostr relays - Letdown2491/trustedrelays

Having an LLM summarize Nostr conversations with little context can actually be hilarious...

I've seen a bunch of "how to lower token usage" posts and YouTube videos lately with overly complicated strategies...here's a simple one: -Use LLMs to help you interpret data, not get data for you. You'll remain a human with a beautiful brain that can think for itself, learn new skills, and make your life easier at the same time.

Just calmly adding LLM stuff into everything I guess...

I think I'll be ready to release this in a week.

Adding a good LLM actually makes this much more useful. For context, an LLM sorts your Nostr timeline into clusters of topics, then writes story summaries based on notes from users you follow so surface interest conversations with network context. Still a long way to go but pretty cool so far...

This new Nostr experience is turning out pretty good...

Not gonna lie...this Nostr client is turning out pretty good....client-side LLM is not nearly powerful enough for good summaries, but might be good enough...

a bot free nostr experience with a user controlled algorithm...a VERY alpha look... this client displays nostr notes in a magazine style and clusters similar notes together using knowledge vectors and a tiny client side llm. the goal is to eventually create "story" summaries of only topics that users will find interesting and with a user-configurable algorithm that updates content in real time. Great, relevant content without the doom-scrolling.