Ralf Roeber (@CognosR) header
Ralf Roeber (@CognosR) profile picture

🎸 local AI agents + FOSS 🖥 mac-stats 🏢 2 companies on autopilot IBM Champion · BCN

Cyberspace & Barcelona, Spain
Joined February 2021
I got tired of explaining my agent loop in replies. So I wrote it down. No agent framework. The loop is bash, folder names, and gh. Local model for triage. Cursor / Claude Code / OpenCode only when there is code to edit. 36 issues from one user on a real POS, closed in that loop. Reviewer → coder → tester → close. This repo is the why: github.com/raro42/agents-exp… To adopt it: copy agents2/ github.com/satisfecho/pos/tr… + pos-cursor-loop.sh from satisfecho/pos. Keep the shell. Swap the coder if you want.
2
1
324
The power and danger of abliterated models. How to build a nuclear bomb, or gunpowder. This is a model I downloaded 11 month ago and totally forgot about it. @support_huihui has only 40 followers on github, 2000+ on X
1
2
89
Abliterated is the future and huihui the god. I finetuned an abliterated model on my WhatsApp chats this weekend. And it really started to answer as if it was me. Short, technical, facts only, no BS. Crazy what you can do if you have the weights.
99
Had to work with Copilot Agent GUI today: click a version and it rolls the chat back to that version. By itself. It did not just scroll the chat. It wrote the old script over the current file. On a shared NAS. A production script. The history is the UI. Not git. That is why I keep the loop outside the IDE. The commit is the version. A click cannot undo it.
1
1
30
Closed firewall by default: yes. Tailscale as the only way in: no. That is their control plane. WireGuard is the protocol. I already run that.
How do I tell the @Hetzner_Online and @digitalocean CEOs to make new VPS servers: - have a default firewall that closes all their ports - by default have Tailscale installed as the only way to access it via SSH - by default have some web tunnel like Cloudflare Tunnel to serve web traffic safely That way VPS servers could get a better reputation instantly by being safe by default It's crazy in 2026 they let you create a VPS that by default is open to the entire internet!
42
I just sponsored sebres for working on fail2ban and nginx with $230. Go sponsor your open source dependencies! github.com/sponsors/sebres?o… Happy birthday @SantoshYadavDev
29
121 commits in 2021 on public repos in github - without AI 18.625 commits in 2026 - with AI Finally I get things done that I want to be done. Above numbers are commits from public repos ... the numbers do not include private repos ;-) For me the productivity gain is incredible and on somedays overwhelming.
1
24
I prefer find/search + BM25F over RAG. My pain points with RAG is finding the right chunk size and updating the RAG-DB. Wonder if there is any better way to retrieve meaningful context from a lot of MD-files to feed into LLM?
1
1
1
34
Am I the only one confusing the remove decimals buttons in google calc? tbh ... in 100% of the cases I first hit the right button and then the left button 3x. Is my brain wiring wrong?
18
I am 7 years on X and have never shared a screenshot. So, pls clean up your algorithms.
1
1
38
Token burn went down. Quality gates got stricter. Laya (system one model) does the triage. OpenCode only edits. Same thin loop. No harness. Why that still works for me: github.com/raro42/agents-exp…
3
96
I am Ralf, Raffa, raro42, One of only 1k IBM champions, probably the only one have decompiled IBM Cognos and submitted a PMR. I built open source, I love to connect, to travel. Let’s connect and do something good for our children.
6
3
99
This pic is real. I am real. No AI involved.
1
14
Things happen and then you meet ppl and then you interact and magic happens. I have not personally talked to Juergen Schmidhuber. But I think it could change many things. Hope to see you soon
1
29
I was lucky to argue with Prof Weizenbaum about the future of AI back in 1990ths Can confirm what you are saying. Thanks 🙏 for sharing.
I have thought about this since the 1970s. See the summary in Sec. 20 of reference [1]: What will supersmart AIs do? As indicated above, they will start by transforming the solar system through countless self-replicating, self-improving robot factories. But that won't be the end of it. In a few hundred thousand years (just a moment on a cosmic scale), they will have transformed the entire galaxy, which has a diameter of about 100,000 light-years. (The last millennium saw many reasonable proposals for interstellar travel, which is much easier for robots than for fragile humans.) Over time, AIs will set up physical senders and receivers everywhere, allowing them to travel by radio, just as they do in our labs today. In a few tens of billions of years, they will have transformed the rest of the reachable universe. Despite the light-speed limit, the expanding AI sphere will have ample time to colonize and shape the entire visible cosmos [TED12][SA17]. Perhaps all of this has already started elsewhere in the universe. Then why don't we see any signs of intelligence out there? (The Fermi paradox.) I've thought about this a lot. When I was a boy, I read about the vast empty spaces observed between clusters of galaxies. My first idea was that they were expanding bubbles colonized by AIs that used most of the local energy, making those bubbles appear dark. Then, I learned that gravity itself is sufficient to explain the sparse, large-scale network structure of the universe. Next, I thought that the mysterious "dark matter," which makes up most of the universe's mass, might be stars whose energy is used by AI civilizations whose communications are so well encrypted that they appear as random noise to us. However, this also seemed implausible because dark matter is present in all galaxies, including our own. So, why are there any stars left in the Milky Way whose energy hasn't been used yet? And why aren't we bombarded with non-encrypted construction plans from AIs that want to spread via radio without first building physical receivers far from their origins? Today, I think it's possible that our planet is the first in our light cone to spawn an expanding AI bubble. Earth's multi-billion-year window for biological evolution is almost over. In a few hundred million years, the sun will be too hot for life as we know it. Perhaps humans were incredibly lucky to evolve in time to invent agriculture, civilization, book printing, and AIs just a few hundred years later. If we are indeed the first, then this implies a great deal of responsibility, not just for our biosphere, but for the future of the entire universe. Let's not mess this up. [1] Annotated History of Modern AI and Deep Learning. Technical Report IDSIA-22-22, IDSIA, Lugano, Switzerland, 2022-2025. Preprint arXiv:2212.11279. people.idsia.ch/~juergen/dee…
2
34
Proud to see satisfecho.de slowly taking off
13
This is happening. And it’s a swarm. It’s happening silently. But, open source will win 🏆 May be I am dreaming.
my spidey senses are telling me the labs are, as we speak, using unguardrailed versions of their unreleased models to systematically reverse-engineer ALL of the internet’s software, apps, and games, then feeding the reconstructed source code straight into their training pipelines while simultaneously building guardrails to ladder-pull the rest of us
1
20
Sad notice. Today we lost our watchdog Blacky amvara.de/de/who-we-are/6 May he rest in peace.
21
Woah. Just sat down at Hotel bar in YUL airport and this fire started. Anybody knows what’s going on in Montreal?
8
3
1,666
@MTL_SIM is on it … xitter.gq/mtl_sim/status/2117001… It’s a resedential building near Atwater. Keep safe.
Feu de bâtiment - Intersection rue Sainte-Cunégonde/ rue de Lévis- Cinquième alarme - Coupure de courant possible. Merci d'éviter le secteur.
126