DevOps’ish

With Talos, most CVEs never apply, and patching the rest won’t break the fleet. Patching a CVE means knowing whether it applies to you, then getting the fix to every node without missing one. Talos Linux ships fewer than 50 binaries, because that’s all it takes to run Kubernetes, so most CVEs never apply. Omni rolls out the ones that do, staged and health-checked, with automatic rollback if a node fails. Designing IaC Interfaces That Work for Humans, AI Agents, and Whatever Comes Next (SPONSOR) AI agents are changing who, or what, uses your Terraform modules. Join Jinger Meilani, Senior DevOps Engineer at MNTN, to learn how to design reusable, self-service IaC interfaces that reduce misuse and work for humans, AI agents, and whatever comes next. Announcing etcd v3.7.0 (11 minute read) SIG etcd ships v3.7.0 with a new RangeStream API for streaming large result sets, keys-only range and faster lease optimizations, plus removal of the legacy v2 store. The quiet workhorse under every Kubernetes cluster gets a meaningful tune-up. ...

Designing IaC Interfaces That Work for Humans, AI Agents, and Whatever Comes Next (SPONSOR) AI agents are changing who, or what, uses your Terraform modules. Join Jinger Meilani, Senior DevOps Engineer at MNTN, to learn how to design reusable, self-service IaC interfaces that reduce misuse and work for humans, AI agents, and whatever comes next. Kepler, re-architected: Improved power accuracy and a community call to action! (8 minute read) The CNCF’s Kubernetes power-monitoring project got a full rewrite. The new architecture drops eBPF, sheds a pile of required privileges, and adds dynamic hardware discovery so the energy numbers actually mean something across mixed fleets. The team is also asking for help validating accuracy, so if you care about sustainability metrics, consider this your invitation. Akrites: The Latest Attempt to Protect Open-Source From AI Attacks Has Arrived (7 minute read) The Linux Foundation stood up Akrites, a single coordination point for finding and fixing open source vulnerabilities before attackers get there first. Jim Zemlin’s framing is bleak and accurate: the mean time to exploit is now measured in negative days. Whether another initiative moves the needle or just adds a logo to the pile is the open question. ...

We helped build Docker. Now we’re building the engineer who maintains it. (SPONSOR) Sam was Docker’s first hire. Andrea wrote Docker’s first commit. We spent a decade watching teams drown in CI maintenance. Mendral is what we wished we’d had. Three agents in your CI: Security reviews dep PRs, Reliability fixes flaky tests, Performance cuts pipeline time. Designing IaC Interfaces That Work for Humans, AI Agents, and Whatever Comes Next (SPONSOR) AI agents are changing who, or what, uses your Terraform modules. Join Jinger Meilani, Senior DevOps Engineer at MNTN, to learn how to design reusable, self-service IaC interfaces that reduce misuse and work for humans, AI agents, and whatever comes next. Klue Supply Chain Incident and LastPass Response (4 minute read) An unauthorized actor snagged OAuth tokens from Klue, a market intelligence platform, and used them to access LastPass customer contact and CRM data stored in Salesforce. LastPass says vaults and core infrastructure are unaffected, but this is another clean example of why your vendor’s vendor is still your problem. ...

A security reviewer on every PR that touches a dependency. (SPONSOR) Most PR reviewers read the diff in your repo. Nobody pulls the source of the dep that just got added or bumped. That’s where supply chain attacks live: a new postinstall script, a network call to a fresh domain, an obfuscated blob. Mendral runs on every PR that touches dependencies as a security-minded reviewer. It checks how recently the version was published, reads the dep’s actual code diff, and flags suspicious patterns before merge. How I’m Solving Local Inference - Running powerful models locally across two laptops using LM Studio’s LM Link, trading token costs for portability between a MacBook Air and Framework 13. Worth a look if you’re tired of the meter running or have a lightweight daily driver and a beefier box somewhere. GitHub dismissed security reports on flaws now exploited by supply-chain worm, researchers say - Deep Specter documented two vulnerability reports GitHub rejected, both of which are now being actively exploited by the Shai-Hulud supply-chain worm, which has already compromised hundreds of packages and developer accounts across major code repositories. ...

The Claude Fable launch turned into a complete debacle this week and is a cautionary tale of the times we’re in. It’s also one of the most botched product releases we’ve witnessed in quite some time. On Tuesday, June 9th, Claude Fable 5 and Mythos 5 were released to every paying customer for a limited time, before it was slated to move to API billing on June 22nd. Anthropic implemented several safeguards (classifiers) to route sensitive topics such as cybersecurity, chemistry, and biology queries to older models. There were additional checks in place that hindered AI researchers and safety folks from doing their work by invisibly doing the same thing. Anthropic walked back the invisible safety checks, saying, “We made the wrong trade-off and we apologize for not getting the balance right.” On Thursday, Microsoft restricted employees from using the latest Claude models due to a 30-day retention policy that Anthropic implemented as yet another safety feature. I have no doubt numerous other organizations would do the same. ...