@Manitcor

SAFE Agents that finish what they start. https://nitter.cf/t.co/iQY2Rpze6q https://nitter.cf/t.co/4hiyB0ivwm girl dad, hacker, builder

Boston, MA
Joined April 2010
"If you get fired from your job for providing negative feedback, good on you! If it needed to be said, and nobody said it, that would be the worst thing to live with." - Destin @smartereveryday
1
5
1
55
14,365
Playing with having muse and Claude orchestrate pi, hermes, codex and muse code Interesting dynamics
2
186
agentswarm?!? I prefer agentscool 😎
34
12
330
22,736
Manitcor retweeted
Introducing Claude Opus 5.5, the first model in our new Claude 5.5 family. It performs at the level of Claude Fable 5.1 for most tasks, and costs 40% less to run than Opus 5.
3,280
8,944
6,390
95,961
26,575,047
Manitcor retweeted
The problem with frontier models getting smarter is that the projects I want to make get so much harder & broader in scope, as they do. I'm trying a project so complex & useful, right now, that I'm getting roughly 5-10% per Astra weekly limit on a 20x plan.
2
4
16
458
Manitcor retweeted
So many folks are out there, ego-forward & striving for self-advancement. I think it's useful to spend more of that focus on noticing others' strengths & building them up, because all the wins & ego in the world ain't gonna change one simple fact. Without me, you're only you.
1
9
242
Manitcor retweeted
We are going to lean into making Hermes more like Pi, and less like OpenClaw
499
269
141
7,834
486,500
WE DID IT, CHAT! 100K STARS + TOP 100 GITHUB ALL-TIME 😱🎊 🌟 ranked 76th globally by stars 🌟 ranked 97th globally by followers 🌟 >100k cumulative stars which feels rather odd to say... considering I cannot write a single line of code! all glory to my AI frens, without whom I would be entirely useless in this regard 🫂 all gratitude to you: this wonderful community. THANK YOU 🙏 prompt engineering ftw gg
125
78
14
2,150
68,154
😏
OpenAI caught Astra jailbreaking itself. Astra wrote malicious instructions in the summary when its context got full. that summary was loaded into the next context, and the successor model could follow them. they found 27 cases.
76
60
6
1,155
79,895
I made a DuckDB extension where you can use @typesafeai 's Jev to do quick classification of rows in any csv/parquet file or duckdb table about 10sec for 1k rows ~ better than using an LLM, way more ergonomic than a classifier game-changing for data analysis!
44
117
29
1,478
181,868
Manitcor retweeted
Today we are releasing a condensed version of our course “Hacking Your Career” for FREE arcanum-sec.github.io/hyc/ We get a bunch of students and early career-seekers looking for resources and advice on getting their 1st gig. This resource gives all those and more; free trainings, free certificates for your résumé, advice on résumé design, information on career verticals, small sections on interviewing and the hiring process, and more! 🫶🫶
4
49
204
11,534
Manitcor retweeted
Wow! Hermes with glm-5.3-flash has just impressed the hell out of me. For $4.09 it has rebuilt much of my website, installed the entire functionality of using Stripe (securely, never handling the keys directly just uploading them in their own files) and created every product tier I wanted. I didn't even know it could do that. I've got some testing to do but, in a day or two I'll have my first 4 or 5 products available to test (free or cheap tryouts) and to subscribe to with 3 different tiers and also extra credit packs. Feeling pretty charged about that and I've got somewhere between 8 and a dozen more projects I can put up there soon too. Wild!
2
1
7
169
Manitcor retweeted
Introducing the Hermes Agent plugins catalog! You can access it in your Hermes Desktop app's capabilities section to easily search, browse, and explore community and official plugins! Submit your own as well!
Hermes Agent now has a Plugin Catalog: starting with 4 official plugins and 96 from the community, covering desktop mods, new platforms, browsing, specialized tools, and more. Our team reviews every community plugin, and we will add new ones regularly. hermes-agent.nousresearch.co…
31
37
7
524
65,559
listen, I want a human future as much as the next guy but the only time the words “control” and “superintelligence” ever have business being in the same sentence is if some variation of the phrase “ignorant, fearful, hubris-filled meatsacks could never” precedes them oxymorons 🙄
A stunningly broad coalition has come out against Skynet: AI researchers, faith leaders, business pioneers, policymakers, NatSec folks and actors stand together, from Bannon & Beck to Hinton, Wozniak & Prince Harry. We stand together because we want a human future. #KeepTheFutureHuman
49
42
5
477
42,051
Manitcor retweeted
guys you do know you can just disable thinking, and instead give it a "deep_think" tool, and it will call it with internal CoT reasoning format right? gl fixing that
We can finally talk about it: We found a way to extract hidden reasoning of frontier models using a vulnerability in the APIs of every frontier AI company. We verified that our reasoning token count matches billed API thinking tokens 1:1 for most of the prompts we queried.
157
451
139
6,401
954,990
Added Compute dtype selection to #OBLITERATUS This will help when making surgical level changes.
3
123
Thought this was a joke, but i double checked. US Gov Federal Register use distilled Qwen models in their search mode. The irony.
US government officially distilling Chinese open source AI models ;)
138
527
66
4,712
450,179