LodeHQ
AI news for people who shipFree, by email

DevOps · Friday, 31 July 2026 · Issue 59 · 9 stories

GitHub stacked PRs, DORA AI findings, K8s 1.37 bans

CI/CD & Automation

DORA report shows AI boosts dev performance only when org systems are tuned

New DORA 2025 research reveals AI is an amplifier, not a magic fix. Teams that invest in foundational processes, clear priorities, robust architecture, and data pipelines, see up to 30% higher delivery speed and stability, while weak systems merely ship faster bugs. The report outlines seven team profiles and capabilities to guide effective AI adoption.

dora.dev ↗1 min read at source

Containers & Orchestration

Kubernetes 1.37 Removes kubectl run -f and Bans Secrets in Static Pods

The preview warns that v1.37 will deprecate “kubectl run -f” as it no longer creates pods from files, and static pods will no longer be allowed to reference Secrets or ConfigMaps, breaking a long‑standing bug. It also flags the upcoming removal of ipvs mode in kube‑proxy and the impending phase‑out of cgroup v1, forcing admins to adapt now.

kubernetes.io ↗6 min read at source

EKS Auto Mode & Karpenter add EFA and placement groups

Amazon EKS now lets Auto Mode and Karpenter node pools use EC2 placement groups and Elastic Fabric Adapter (EFA) networking. This gives teams fine‑grained control over instance placement and IP usage, unlocking higher throughput for distributed training and tighter fault isolation for critical services. Available in all regions.

aws.amazon.com ↗1 min read at source

DevSecOps

Cloudflare adds post‑quantum signatures to origin pulls, locking down server‑to‑server links

Cloudflare’s Authenticated Origin Pulls and Custom Origin Trust Store now accept ML‑DSA post‑quantum signatures, enabling fully quantum‑resistant TLS authentication between Cloudflare edge nodes and customer origin servers. This thwarts future quantum impersonation attacks on the server‑to‑origin link, completing the first milestone of Cloudflare’s 2029 post‑quantum security roadmap.

blog.cloudflare.com ↗12 min read at source

CISA's 2026 SBOM Rules Require Component Hashes and AI Model Disclosure

The new CISA guidance replaces the 2021 baseline and makes cryptographic hashes mandatory for every SBOM component, letting teams verify the exact bytes deployed. It also expands the minimum elements to cover AI/ML model artifacts, pushing vendors to disclose model and dataset provenance. Together these changes raise the bar for software supply‑chain transparency and risk management.

cisa.gov ↗2 min read at source

GitLab’s governance playbook for agentic AI code assistants

GitLab warns that agentic AI can bypass human review, letting code changes be made automatically, and offers a concrete governance framework, identity, permissions, audit trails, and an MCP‑enabled checklist, to keep the pipeline secure. The approach targets the 92% of DevSecOps teams already seeing AI‑generated risk, turning control points into a built‑in safety net.

about.gitlab.com ↗7 min read at source

You’re caught up.

That was all 9 stories in the 31 July issue.