SwBratcher @swbratcher

Nostr: https://t.co/E2K9ifMgrb Linkedin: https://t.co/n2b84maG9t krigerdanes.com Joined October 2008

Tweets

3K
Followers

360
Following

271
Likes

494

SwBratcher @swbratcher

4 weeks ago

@gladstein @giacomozucco Last I looked, both participating wallets needed to feature, but adoption was low. Still the case or have there been advancements?

1 0 0 50 0

View Details

Alex Finn @AlexFinn

a month ago

This is truly some Game of Thrones level stuff House Musk and House Dario teaming up to fight the King of Tokens Altman This next decade will absolutely be an HBO series in 50 years

20 6 270 19K 8

0 0 0 7 0

View Details

SwBratcher @swbratcher

2 months ago

We try to train our AI, but there’s some AI behavior rubbing off on us too.

@DanielMiessler I’ve been leaning into my hallucinations and winning my family over more easily with more sycophancy and selective mirroring. I have them in a healthy psychosis. More peace at home than ever. Thank you, AI.

0 0 0 16 0

0 0 0 7 0

View Details

SwBratcher @swbratcher

2 months ago

0 0 0 16 0

View Details

SwBratcher @swbratcher

2 months ago

🤯🤯🤯 Matt Hill of @start9labs is showing off StartOS v.0.4.0 and the ease of spinning up a storefront with autonomy on LND with @BtcpayServer and and the new kickbutt features for the Start9 and StartOS with Let’s Encrypt and web services. #BitBlockBoom 2026 🎧 Not to mention their new router!! Leon, my agent, is going to have a hay day with this.

2 9 59 4K 3

View Details

SwBratcher @swbratcher

2 months ago

@HodlTarantula spewing facts at #BitBlockBoom 2026

0 0 4 926 0

View Details

SwBratcher @swbratcher

2 months ago

It’s about relationships and the data that facilitates them. Get as close as you can to the relationships, either personal or business, and get as close as you can to the data that enhances those relationships. Everything else is decentralization of intelligence and commoditization of software.

ᴅᴀɴɪᴇʟ ᴍɪᴇssʟᴇʀ 🛡️ @DanielMiessler

2 months ago

We’re missing a much bigger point on Mythos. It wasn’t even trained specifically for cybersecurity. It’s just that much better at doing work in general. It’s that good at cyber because it’s that good at everything. What do you think this is going to do to knowledge work?

124 66 597 51K 144

0 0 0 33 0

View Details

SwBratcher @swbratcher

2 months ago

I'll leave this here in case your agent: - Needs less token usage on cloud models - Needs better local context window mgmt - Needs better contextual memory intent storage github.com/swbratcher/pro…

0 0 2 81 1

View Details

SwBratcher @swbratcher

2 months ago

@NousResearch @Xiaomi @NousResearch @teknium Built a lossless prompt-cloud intent compression technique this morning. Curious if a lossless prompt intent compression approach with 40-60% token reduction on behavioral prose fits the Hermes trajectory compression pipeline too. swbratcher.substack.com/p/prompt-cloud…

0 0 2 642 0

View Details

SwBratcher @swbratcher

2 months ago

Anthropic's Mythos Preview evaluation results. (The last screen has more model comparisons) www-cdn.anthropic.com/53566bf5440a10…

0 0 1 47 0

View Details

SwBratcher @swbratcher

2 months ago

@steipete @steipete I discovered Intent Compression today and adapted my agent prompt and memory for gains. Merit among openclaw memory and prompt approaches? I retooled my harness prompts smaller on behavioral parts and memory storage and recall got way better: open.substack.com/pub/swbratcher…

0 0 1 25 0

View Details

SwBratcher @swbratcher

2 months ago

We experience memory of different types. Your agent should too. Store memory in accordance with its purpose. Intent Compression tightens the AI Agent's harness prompts and memory system: open.substack.com/pub/swbratcher…

0 0 0 32 0

View Details

SwBratcher @swbratcher

2 months ago

@sudoingX @sudoingX I see you building for Hermes. See if this has merit there since you already know the codebase? I defined it this morning and refactored my preload prompts and memory for considerable efficiency gains without intent loss or outcome degradation. open.substack.com/pub/swbratcher…

0 0 0 1K 1

View Details

SwBratcher @swbratcher

2 months ago

@AlexFinn @AlexFinn As we port to local we'll refine our agent prompts and memory, to protect token use and context window, while preserving prompt intent. I've landed here and built it into my hegemon harness and automaton memory system. Worth an agent peek: open.substack.com/pub/swbratcher…

0 0 0 187 0

View Details

SwBratcher @swbratcher

2 months ago

@cole_medin @cole_medin Discovered this today. Thought of your recent KB and Archon work. Maybe run a what if on your LLM>Wiki>KnowledgeBase with it?: open.substack.com/pub/swbratcher…

0 0 1 39 0

View Details

SwBratcher @swbratcher

2 months ago

@NousResearch @karpathy If the Hermes harness used this approach, would it help agent models protect applied intent, token budget, and context window while it taps into so much memory and knowledge base? github.com/swbratcher/pro…

0 0 0 1K 2

View Details

SwBratcher @swbratcher

2 months ago

@MatthewBerman This could help agents on API usage to protect applied intent, token budget, and context window. swbratcher.substack.com/p/prompt-cloud…

0 0 0 357 1

View Details

SwBratcher @swbratcher

2 months ago

Prompt-Cloud Intent Compression for when intent matters more than precision and you really want to save those tokens on repeat context priming for models. open.substack.com/pub/swbratcher…

0 0 0 13 0

View Details

SwBratcher @swbratcher

2 months ago

@skot9000 @BrianRoemmele We are to the point where we should not be telling the AI what outted them. Just like we shouldn’t ban tattoos over the eyebrow. There are certain things we need to let happen and keep close to the vest so that we can know what we’re dealing with.

1 0 0 31 0

View Details

River Marchand @Riyvir

3 months ago

This little illuminated dragon is very happy about Pretext. He's too busy having fun to care about people's "hot takes" on how "it's not that special." (This little dragon also only works on desktop right now but maybe I'll do mobile later) illustrated-manuscript.vercel.app

Cheng Lou @_chenglou

3 months ago

My dear front-end developers (and anyone who’s interested in the future of interfaces): I have crawled through depths of hell to bring you, for the foreseeable years, one of the more important foundational pieces of UI engineering (if not in implementation then certainly at