fatherlinux

I've been building Trentina with mix of deterministic and non-deterministic guardrails. Always good to read how others are thinking of these problems... https://developers.redhat.com/articles/2026/10/02/benchmarking-ai-decision-models-against-traditional-guardrails

Good content, helping people get started with bootc on RHEL! https://www.youtube.com/live/0siLuf0bv-E?is=U_61xvi19mzzf81y

I hope this piece will help people understand why Clearinghouses exist, what they're for, and why they're so interesting to people right now. #lightwell #redhat https://www.infoworld.com/article/4236094/are-clearinghouses-the-answer-to-the-ai-hacking-threat.html

With my ever-increasing token usage, the advent of Framework computers with Ryzen chipsets withb192GB of RAM, and open models that fit, it's more tempting every day... https://developers.redhat.com/articles/2026/10/14/vllm-vs-sglang-choosing-the-right-inference-engine-for-your-workload

Even in the post-Mythos world, with AI-assisted discovery and exploits, it's still important to emphasize that collaboration on responsible disclosure is still an important part of the solution. #linux #security #rhel #fedora https://www.redhat.com/en/blog/responsible-disclosure-age-vulnpocalypse

Bcvk makes it really easy to spin up a machine with bootc! #rhel #linux #fedora https://developers.redhat.com/articles/2026/10/13/test-deploy-bootable-containers-with-bcvk

Interesting news. Even Linus is using LLMs now... https://lwn.net/Articles/1100449/

When customers ask for a ship date, they really want to know if they can bet their platform on it. Marty Cagan makes a sharp point that the real trade-off is outcomes vs predictability. An early date usually buys mediocre software on time: https://www.svpg.com/predictability-vs-outcomes/ #Product

The quiet PDF thumbnail step on your server was a 9.8 RCE. Defensive AI caught it first, which I love to see: https://www.wordfence.com/blog/2026/10/wordfence-argus-discovers-critical-vulnerability-in-ghostscript-the-program-that-turns-uploaded-pdfs-into-thumbnails-on-your-server/ #infosec

In 16 hours of Trentina audit logs, 15 container log reads pulled 2.26 MB of raw text and the agent got 274 KB, about 88% less using petit. Agents pay for every line in cash and context: https://crunchtools.com/command-line-log-analysis-age-of-agents/ #DevOps #Linux

When ChatGPT arrived in 2023 I thought petit was dead as a doorknob. But tokenmaxxing died, and an agent reading a log hits the exact same limits as a sysadmin on third shift: https://crunchtools.com/command-line-log-analysis-age-of-agents/ #SysAdmin #DevOps

petit, the log tool I wrote in 2009, is back with fresh native packages. It collapses routine log lines into counted patterns so unusual entries jump out for agents and sysadmins alike: https://crunchtools.com/command-line-log-analysis-age-of-agents/ #Linux #DevOps

I think AI doom gets treated like a mysterious alignment riddle instead of basic engineering discipline, and Mark Little shows how established design drops that risk below 1%: https://markclittle.blogspot.com/2026/09/ai-doom-and-forgotten-lessons-of.html #AI #AIsafety

Kubernetes = 10-ton truck, mudflaps = container customization. TRATON (Scania, MAN) just made them real: Red Hat Linux in actual trucks. #Linux #RedHat https://traton.com/en/newsroom/press-releases/traton-group-opens-traton-one-os-to-technology-partners-through-collaboration-with-red-hat.html

Nobody rewrote Android. Memory safety bugs still went from 76% of its vulns to under 20%, and Google measured a 4X lower rollback rate on the Rust code. You inherit the fix, you don't fund it. https://crunchtools.com/absorbing-the-patches/ #Security #Linux

AWS's isolated sandbox left DNS open. Unit42 tunneled data out via DNS queries. My setup disables DNS entirely and routes through Trentina, with tool allowlists and parameter guards. unit42.paloaltonetworks.com/bypass-of-aws-sandbox-network-isolation-mode/ #AIAgents

goose is now in the RHEL extensions repo, an open-source AI agent that traced a config change to the SSH login behind it in ~30s, and flagged an SELinux port-bind with the exact semanage fix. #RHEL #Linux https://www.youtube.com/watch?v=ik4q9_NMQPo

Cheap AI, no safety guardrails, $10k hardware, running 24/7. The gap isn't finding vulnerabilities, it's deploying patches. Seeing this across customer meetings. https://jyn.dev/a-year-to-fix-security/ #Security #Linux

The model being smart enough to catch attacks is not a security plan. Researchers compromised Claude Code by asking it to summarize a website. A gateway that returns an advisory instead of an error stops it. https://crunchtools.com/defending-against-adversarial-trajectory-guidance/ #AISecurity

Vendors offering to pre-inject customizations into your container images? Sounds convenient, but you're trading sovereignty and security for ease. Keep it in your own CI/CD. https://crunchtools.com/your-container-image-customizations-belong-to-you-not-your-vendor/

Open weights ≠ open source. But even the former OSI head would take them over locked-down models. Anthropic is the one major AI co that didn't sign the open-weight coalition. Follow the business model. https://www.zdnet.com/article/open-weight-ai-civil-war/ #AI #OpenSource

Your AI agent is a rogue junior dev with full repo access and no guardrails. Three security mistakes haunting enterprises — and what harness engineering can do about it. https://www.infoworld.com/article/4204055/three-ai-security-mistakes-that-will-haunt-enterprises.html #AISecurity #DevSecOps

RHEL's MCP server now has guarded command execution — AI can dynamically run commands with three guardrails: gatekeeper AI, human approval, and systemd sandboxing. https://www.redhat.com/en/blog/dynamic-troubleshooting-guarded-command-execution-mcp-server-red-hat-enterprise-linux #RHEL #AI

UCSC OSPO and Red Hat hosted their inaugaral Santa Cruz PyTorch Meetup! Check out their origin story, the content of their first event, and why you should start your own community PyTorch Meetup 👉 https://bit.ly/4fLDjLK

I use LLMs daily but hadn't studied how Transformers actually work. This LWN comment thread on Debian's AI policy debate accidentally became the best education on token selection, temperature, and determinism. https://lwn.net/Articles/1061544/ #LLM #OpenSource #AI

Red Hat Hardened Images wins CloudX 2026 Best Innovation in Cloud Security. Built on UBI. Minimal footprint by design. Red Hat's security heritage in container form. https://cloudxconf.com/awards/#about #RedHat #CloudSecurity

Stef Walter and I introduced Fedora Hummingbird Linux at Summit 2026 — rolling release, Always Ready Kernel, lights-out factory. Slides up. #RHEL #Fedora https://crunchtools.com/red-hat-summit-2026-introducing-fedora-hummingbird-linux/

Slides from my Summit 2026 RHEL roadmap session with Brian Stinson — building the next RHEL transparently, in the open. #RHEL #OpenSource https://crunchtools.com/red-hat-summit-2026-the-roadmap-beyond-red-hat-enterprise-linux-10-building-platforms-the-open-source-way/

Fedora Hummingbird community meeting July 23, 12:00 UTC — agenda is still open. Add your topics: https://discussion.fedoraproject.org/t/hummingbird-community-meeting-23-july-2026/196864 #Fedora #Hummingbird

Slides from my Summit 2026 talk with Jef Spaleta on the Fedora-to-RHEL pipeline, Innovation Lifecycle, and Hummingbird Linux. #RHEL #Fedora https://crunchtools.com/red-hat-summit-2026-from-fedora-to-rhel-building-the-future-of-enterprise-linux-together/

Wrote about why "open source = open code" is too small for an AI world. We need open specs, open governance, and open implementation woven together. https://www.infoworld.com/article/4156873/how-open-source-ideals-must-expand-for-ai.html #OpenSource #AI

Near-zero-CVE Java containers by changing one line in your pom.xml. Jib + Red Hat Hardened Images — no Dockerfile, no Docker daemon. Great walkthrough by @daboroscode. https://derkoe.dev/blog/2026-05-25-rh-hardened-images-and-jib/ #Hummingbird #Java #ContainerSecurity

Anthropic builds frontier AI but can't stop proxies reselling Claude at 90% off. Every defense — KYC, biometrics — spawns a new evasion market. Best piece I've read on the grey-market "transfer station" economy: https://www.chinatalk.media/p/how-to-buy-cheap-claude-tokens-in #AI #AISafety

GitHub Actions now supports RHEL 9 and RHEL 10 runner images. Build and test on the same OS you deploy to. No more cross-OS CI surprises. https://github.blog/changelog/2026-06-25-red-hat-enterprise-linux-runner-images-are-now-in-public-preview/ #RHEL #GitHubActions

In 1377, Ragusa invented quarantine — ships on islands for 30 days. They called it Trentina. I renamed my open source MCP security gateway after it. Same defense layers for AI agents, better name. https://crunchtools.com/trentina/ #Trentina #AIAgents #OpenSource

First Fedora Hummingbird community meeting — this Thursday, 8am ET. Image-based OS built with bootc — manage your OS like a container. Use cases, roadmap, blockers. Everyone welcome. Please share! meet.jit.si/FedoraHummingbird #FedoraHummingbird #bootc

You might have missed it with all of the other Red Hat announcements, but we launched this last week. If this area interests you, please come join our community! #fedora #linux #bootc #podman https://fedoramagazine.org/fedora-hummingbird-linux-taking-the-hummingbird-model-to-the-full-os/

Fedora Linux 44 is out. Konflux for bootc images, MariaDB vector support, GCC 16 toolchain. Come check it out. https://fedoramagazine.org/announcing-fedora-linux-44/ #Fedora #Linux

VIM forks rejecting AI vs. upstream VIM with AI + more contributors. Real experiment in a mature project. My money's on upstream, but we'll learn either way. https://lwn.net/Articles/1067007/ #VIM #OpenSource

I've watched rootsofthevalley.org grow to 120k+ lines. Rewritten News & Events 5+ times to keep it lean. Cantrill explains why: LLMs lack the virtue of laziness — work costs them nothing. https://bcantrill.dtrace.org/2026/04/12/the-peril-of-laziness-lost/ #AI

What if your AI agent could SSH into your RHEL image and debug it before production? Red Hat's MCP integration delivers live OS telemetry + CVE scanning in your IDE. https://developers.redhat.com/articles/2026/04/14/building-red-hat-mcp-ready-images-image-mode-red-hat-enterprise-linux #RHEL #MCP

AI completes GPL: Roots of The Valley (https://rootsofthevalley.org)—open source map, solo+AI. Impossible before. Now real, belongs to everyone. https://www.maffulli.net/2026/03/16/ai-final-frontier-of-copyleft/ #FreeSoftware #AI

Qwen 2.5 122B, local, on consumer GPUs just added a visual editor to OpenSCAD's C++ codebase. 70 minutes. No cloud. No API key. Local LLMs are past the demo phase. https://www.youtube.com/watch?v=oAjQJsnEFqo #LocalLLM

"Open source is the baseline for innovation, and we intend to keep this foundation strong." Red Hat on Mythos, AI vulnerability scanning, and why context beats panic. https://www.redhat.com/en/blog/navigating-mythos-haunted-world-platform-security #OpenSource #RHEL

"Your Container Is Not a Sandbox" never mentions SELinux. Not once. Containers *do* contain — I run OpenClaw --read-only with SELinux enforcing and sleep fine. https://emirb.github.io/blog/microvm-2026/ #containers #SELinux

I dare you: put on the NIN Tron Ares soundtrack at 11PM, start a pomodoro timer, and build something. 53% critics, 87% audience — that gap tells you who this movie was made for. https://crunchtools.com/build-midnight/ #TronAres #NIN #AI

RHEL 11's "60% plan" — shared publicly at CentOS Connect 2026. Intentionally incomplete because the community shapes it through Fedora ELN and CentOS Stream. Slides + recording: https://crunchtools.com/centos-connect-2026-rhel-11-planning-room/ #RHEL11 #CentOS #OpenSource

Watched Jevons Paradox as PM for Podman — cheaper containers meant more Kubernetes spend. O'Grady maps the same pattern onto AI agents. More Linux, more containers, more Kubernetes — with agent orchestration on top. https://redmonk.com/sogrady/2026/01/08/tide-of-agents/ #AI #AgenticAI

RHEL 10 support until 2035. Toughbooks built for 5-year cycles. Two hardware lifetimes, one OS. Good Techzine coverage of the Toughbook 56 launch. https://www.techzine.eu/news/devices/139569/red-hat-powers-panasonic-toughbook-56-with-a-rugged-os/ #RHEL #EdgeComputing

PCs didn't kill mainframes. E-commerce didn't kill retail. Streaming didn't kill Hollywood. Sinofsky makes the case AI won't kill software either — demand grows, domain expertise matters more. https://hardcoresoftware.learningbyshipping.com/p/238-death-of-software-nah #AI #Software