This is a guest conversation episode of Ship It Weekly, separate from the weekly news recaps.

In this Ship It: Conversations episode, I talk with Stephane Moser about Pipedrive’s move from Jenkins to GitHub Actions, building self-hosted runners on Kubernetes, shifting deployments toward GitOps with Argo CD, and what it actually takes to roll out a big CI/CD change across a large engineering org.

We talk about why Jenkins had become painful, from Groovy friction to noisy-neighbor problems on shared VMs, why GitHub Actions fit better, how reusable workflows and custom actions helped, why Argo CD beat out Flux for their use case, and how they had to build better observability and internal deployment visibility around GitHub as they scaled.

The bigger theme here is that this was not just a tooling swap. It was a product and platform migration. Isolation, repeatability, self-service, rollout strategy, and observability mattered just as much as the actual CI/CD tools.

Highlights

• Why Jenkins stopped working well for them: Groovy friction, shared VM contention, and poor predictability

• Replacing CodeShip pull request validation first as the low-blast-radius starting point

• Using Actions Runner Controller on Kubernetes with EKS and Karpenter for self-hosted runners

• Why reusable workflows and custom actions helped cut repetition across hundreds of services

• Choosing Argo CD over Flux, Argo Workflows, Tekton, and even a short Spinnaker attempt

• Moving from push-based deploys toward GitOps for better isolation and safer credentials handling

• Building internal observability because GitHub’s workflow visibility was not enough at their scale

• Dogfooding first, then rolling migration out in batches until teams could self-serve the move

• What broke when the new system actually worked too well: bot-driven deploy volume, queueing, and fairness

• The mobile side of the story: Mac minis, unstable runners, GitHub-hosted runners, and a very different migration path

• How AI sped up parts of the mobile migration and troubleshooting, without making the migration trivial

• Stephane’s advice for big CI/CD shifts: start small, reduce blast radius, and use your own platform first

Stephane’s links

• LinkedIn: https://www.linkedin.com/in/moserss/

• Talk video: https://www.youtube.com/watch?v=VrE1dh-1zEY

• Blog post Part 1: https://medium.com/pipedrive-engineering/so-long-jenkins-hello-github-actions-pipedrives-big-ci-cd-switch-03be29c75f63

• Blog post Part 2: https://medium.com/pipedrive-engineering/all-aboard-the-github-actions-express-pipedrives-big-ci-cd-switch-part-2-fcacf834afd2

• GitHub: https://github.com/moser-ss

Our links

More episodes + show notes + links: https://shipitweekly.fm

On Call Brief: https://oncallbrief.com

This episode of Ship It Weekly is about networking, ingress, and private access moving further up into the platform layer. Brian covers AWS Interconnect going generally available, Cloudflare Mesh, GitLab 19.0 breaking changes around Gateway API and bundled services, EKS Auto Mode networking, and OpenTelemetry declarative config reaching stability. He also hits containerd security patches, GitHub’s new Code Security risk assessment, and AWS guidance on securing AI agents with MCP. (Amazon Web Services, Inc.)

Links

AWS Interconnect GA and last mile connectivity https://aws.amazon.com/blogs/aws/aws-interconnect-is-now-generally-available-with-a-new-option-to-simplify-last-mile-connectivity/

Cloudflare Mesh https://blog.cloudflare.com/mesh/

GitLab 19.0 breaking changes https://about.gitlab.com/blog/a-guide-to-the-breaking-changes-in-gitlab-19-0/

EKS Auto Mode networking https://aws.amazon.com/blogs/containers/navigating-enterprise-networking-challenges-with-amazon-eks-auto-mode/

OpenTelemetry declarative config reaches stability https://opentelemetry.io/blog/2026/stable-declarative-config/

containerd security releases https://github.com/containerd/containerd/releases

GitHub Code Security risk assessment for organizations https://github.blog/changelog/2026-04-08-code-security-risk-assessment-available-for-organizations/

AWS secure AI agent access patterns using MCP https://aws.amazon.com/blogs/security/secure-ai-agent-access-patterns-to-aws-resources-using-model-context-protocol/

This week’s On Call Brief https://www.tellerstech.com/on-call-brief/2026-W16/

More episodes and show notes https://shipitweekly.fm/

In this Ship It Weekly special, Brian breaks down Claude Mythos Preview and Project Glasswing, and why this story matters beyond normal AI launch hype.

Anthropic is treating Mythos like a real security inflection point, not just a better coding model. Project Glasswing is their coordinated effort to get early access into the hands of defenders, critical software maintainers, and major infrastructure organizations before similar capability becomes more broadly available. If OpenClaw was about agents becoming a new control plane, this episode is about what happens when finding ways into messy environments and control planes starts getting faster too.

We walk through the practical angle for DevOps, cloud, platform, and infra teams: exploit timelines may be compressing, platform debt becomes attacker leverage, and the boring work most orgs treat like cleanup suddenly looks a lot more like frontline security work. We also zoom out to the business side, including why banks, regulators, and government officials are already paying attention.

Chapters

• Why This Episode Exists
• OpenClaw Callback
• What Actually Happened
• Don’t Get Gullible, Don’t Get Lazy
• What Changes If This Is Even Half True
• Why Business People Should Care
• What This Means for DevOps, Cloud, and Platform
• Boring Work Just Got Promoted
• The Uncomfortable Takeaway
• What I’d Do Right Now

Links from this episode

Claude Mythos Preview

https://red.anthropic.com/2026/mythos-preview/

Project Glasswing

https://www.anthropic.com/project/glasswing

AI cyber threats: open letter to business leaders

https://www.gov.uk/government/publications/ai-cyber-threats-open-letter-to-business-leaders/ai-cyber-threats-open-letter-to-business-leaders-html

AI-boosted hacks with Anthropic’s Mythos could have dire consequences for banks

https://www.reuters.com/legal/litigation/ai-boosted-hacks-with-anthropics-mythos-could-have-dire-consequences-banks-2026-04-13/

ECB to quiz bankers about risks of Anthropic's new AI model, source says

https://www.reuters.com/world/ecb-warn-bankers-about-new-anthropic-model-risks-source-says-2026-04-15/

Related episode: OpenClaw special

https://www.tellerstech.com/ship-it-weekly/special-openclaw-security-timeline-and-fallout-cve-2026-25253-one-click-token-leak-malicious-clawhub-skills-exposed-agent-control-panels-and-why-local-ai-agents-are-a-new-devops-sre-control-plane/

This episode of Ship It Weekly is about the interface layer becoming the story. Brian covers Amazon S3 Files and why it feels more like a managed filesystem layer in front of S3 than “S3 is EFS now,” including how it relates to the old s3fs and FUSE-style approach. He also digs into 36 malicious npm packages posing as Strapi plugins, the uglier follow-on to the Trivy incident he discussed previously, Kubernetes Ingress2Gateway 1.0 and the push toward Gateway API, and Kubernetes Agent Sandbox as a sign that newer AI-style workloads are starting to reshape the platform itself.

Links

Amazon S3 Files

https://aws.amazon.com/blogs/aws/launching-s3-files-making-s3-buckets-accessible-as-file-systems/

Malicious npm packages posing as Strapi plugins

https://thehackernews.com/2026/04/36-malicious-npm-packages-exploited.html

Trivy follow-on incident discussion

https://github.com/aquasecurity/trivy/discussions/10425

RoseSecurity on Trivy / typosquatting angle

https://rosesecurity.dev/2026/03/20/typosquatting-trivy.html

Earlier episode covering the first Trivy incident

https://www.tellerstech.com/ship-it-weekly/aws-bahrain-uae-data-center-issues-amid-iran-strikes-argocd-vs-flux-gitops-failures-github-actions-hackerbot-claw-attacks-trivy-roguepilot-codespaces-prompt-injection-block-ai-remake/

Kubernetes Ingress2Gateway 1.0

https://kubernetes.io/blog/2026/03/20/ingress2gateway-1-0-release/

Kubernetes Agent Sandbox

https://kubernetes.io/blog/2026/03/20/running-agents-on-kubernetes-with-agent-sandbox/

Fortinet FortiClient EMS emergency patch

https://www.fortiguard.com/psirt/FG-IR-26-099

Karpathy post

https://x.com/karpathy/status/2036487306585268612

ProofShot

https://github.com/AmElmo/proofshot

More episodes and show notes

https://shipitweekly.fm

On Call Briefs

https://oncallbrief.com

This is a guest conversation episode of Ship It Weekly, separate from the weekly news recaps.

In this Ship It: Conversations episode, I talk with David Chute, founder and CEO of Roadie, about internal developer portals, Backstage, automation, and how IDPs may evolve as AI agents become more common in engineering workflows.

We talk about the difference between a platform and a portal, the three common problems IDPs usually try to solve, why discoverability tends to be the first pain teams feel, and why a lot of orgs should start with automation before trying to perfect a service catalog. We also get into self-hosted Backstage vs managed options, and how teams should think about adoption, data models, and time to value.

The bigger theme is the one I found most interesting: IDPs may be shifting away from dashboard-heavy “single pane of glass” thinking and toward becoming context layers for workflows, terminals, and eventually agents.

Highlights

• The difference between an internal developer platform and an internal developer portal

• The three common IDP problem areas: discoverability, automation, and guardrails

• Why discoverability is usually the first pain teams feel

• Why adoption is often more of a human problem than a technical one

• Catalog completeness vs team ownership

• Why a lot of teams should start with automation first

• Self-hosted Backstage vs SaaS tradeoffs: extensibility, control, lock-in, and time to value

• Why IDPs may move from dashboards to context delivery for humans and agents

• Why AI helps teams build faster, but does not solve the problem of building the right thing

• David’s advice for platform and DevEx teams: talk to your internal users first

David’s links

• LinkedIn: https://www.linkedin.com/in/davidtuite/

Roadie / Backstage

• Roadie: https://roadie.io/

• Backstage: https://backstage.io/

Stuff mentioned

• Workday

• Backstage

• GitHub

• GitLab

• Bitbucket

• Azure DevOps

• Argo CD

• LaunchDarkly

• CircleCI

• DORA metrics

• MCP-style context for agents

Our links

More episodes + show notes + links: https://shipitweekly.fm

On Call Brief: https://oncallbrief.com

This episode of Ship It Weekly is about the quiet platform work that keeps things safe before they break. Brian covers GitHub Actions hardening in Kubernetes-related repos, Airbnb’s safer config rollouts, Cloudflare’s zero-downtime Rust restarts, Amazon ECS Managed Daemons, and HCP Terraform access controls with IP allow lists and temporary AWS permission delegation.

Links

GitHub Actions security roadmap

https://github.blog/news-insights/product-news/whats-coming-to-our-github-actions-2026-security-roadmap/

Airbnb config rollouts

https://medium.com/airbnb-engineering/safeguarding-dynamic-configuration-changes-at-scale-5aca5222ed68

Cloudflare graceful restarts for Rust

https://blog.cloudflare.com/ecdysis-rust-graceful-restarts/

Amazon ECS Managed Daemons

https://aws.amazon.com/about-aws/whats-new/2026/04/amazon-ecs-managed-daemons/

HCP Terraform IP allow lists

https://www.hashicorp.com/blog/hcp-terraform-adds-ip-allow-list-for-terraform-resources

HCP Terraform AWS permission delegation

https://www.hashicorp.com/blog/aws-permission-delegation-now-generally-available-in-hcp-terraform

GitHub secret scanning updates

https://github.blog/changelog/2026-03-10-secret-scanning-pattern-updates-march-2026/

GitHub secret scanning for AI coding agents

https://github.blog/changelog/2026-03-31-secret-scanning-extends-to-ai-coding-agents-via-the-github-mcp-server/

Codespaces GA with data residency

https://github.blog/changelog/2026-04-01-codespaces-is-now-generally-available-for-github-enterprise-with-data-residency

Kubernetes v1.36 sneak peek

https://kubernetes.io/blog/2026/03/30/kubernetes-v1-36-sneak-peek/

GKE Inference Gateway

https://cloud.google.com/kubernetes-engine/docs/concepts/about-gke-inference-gateway

More episodes and show notes

https://shipitweekly.fm

On Call Briefs

https://oncallbrief.com

This episode of Ship It Weekly is about the places where convenience quietly turns into trust.

Brian revisits the Trivy story by zooming out to the bigger hackerbot-claw GitHub Actions campaign, then gets into the Xygeni tag-poisoning compromise, GitHub’s search high availability rebuild for GitHub Enterprise Server, Windows Server 2025 surfacing duplicate SID problems in cloned images, and the agent-skills ecosystem replaying package supply chain history. Plus: a quick lightning round on GitHub pausing self-hosted runner minimum-version enforcement and March secret scanning updates.

Links

OpenSSF advisory on active GitHub Actions exploitation https://seclists.org/oss-sec/2026/q1/246

Xygeni action compromise via tag poisoning https://www.stepsecurity.io/blog/xygeni-action-compromised-c2-reverse-shell-backdoor-injected-via-tag-poisoning

GitHub Enterprise Server search high availability rebuild https://github.blog/engineering/architecture-optimization/how-we-rebuilt-the-search-architecture-for-high-availability-in-github-enterprise-server/

Microsoft on duplicate SIDs and nongeneralized Windows Server 2025 images https://learn.microsoft.com/en-us/troubleshoot/exchange/administration/exchange-server-issues-on-incorrect-windows-server-image

Socket on supply chain security for skills.sh https://socket.dev/blog/socket-brings-supply-chain-security-to-skills

Snyk ToxicSkills research https://snyk.io/blog/toxicskills-malicious-ai-agent-skills-clawhub/

GitHub self-hosted runner minimum version enforcement paused https://github.blog/changelog/2026-03-13-self-hosted-runner-minimum-version-enforcement-paused/

GitHub secret scanning pattern updates, March 2026 https://github.blog/changelog/2026-03-10-secret-scanning-pattern-updates-march-2026/

More episodes and show notes at https://shipitweekly.fm

On Call Briefs at https://oncallbrief.com

This is a guest conversation episode of Ship It Weekly, separate from the weekly news recaps.

In this Ship It: Conversations episode, I talk with Ang Chen from the University of Michigan about Project Vera, a cloud emulator built to help teams test infrastructure changes more safely before they touch real cloud.

We talk about why testing against real cloud APIs is slow, expensive, and risky, how Vera works under tools like Terraform and CloudFormation, what “high fidelity” actually means, and where a tool like this could fit in local dev and CI/CD.

The bigger theme is one I think matters a lot: if AI is going to play a real role in cloud operations, it probably needs a sandbox first, not direct access to production.

Note

This interview was recorded on February 13, 2026. Since then, Vera’s public project materials have expanded the framing a bit further around multi-cloud support and safe environments for agent learning, so keep that in mind while listening.

Highlights

• Why real cloud testing still creates cost, delay, and risk

• How Vera emulates cloud behavior at the API layer

• Where this could help with Terraform, CloudFormation, and CI/CD workflows

• Why “useful enough to catch real mistakes” may matter more than perfect emulation

• The limits, tradeoffs, and fidelity questions that still need to be solved

• Why safe training grounds may matter before AI agents touch real infrastructure

Ang’s links

• LinkedIn: https://www.linkedin.com/in/ang-chen-8b877a17/

• University of Michigan profile: https://eecs.engin.umich.edu/people/chen-ang/

• Publications: https://web.eecs.umich.edu/~chenang/pubs.html

Project Vera

• Project site: https://project-vera.github.io/

• GitHub: https://github.com/project-vera/vera

• The quest for AI Agents as DevOps: https://project-vera.github.io/blogs/cloudagent/cloudagent/

• No More Manual Mocks: https://project-vera.github.io/blogs/cloudemu/cloudemu/

Stuff mentioned

• A Case for Learned Cloud Emulators: https://dl.acm.org/doi/10.1145/3718958.3754799

• Cloud Infrastructure Management in the Age of AI Agents: https://dl.acm.org/doi/abs/10.1145/3759441.3759443

• LocalStack: https://www.localstack.cloud/

Our links

More episodes + show notes + links: https://shipitweekly.fm

On Call Brief: https://oncallbrief.com

This week on Ship It Weekly, Brian looks at what happens when new interfaces create old responsibilities.

McKinsey patched a vulnerability in its internal AI tool Lilli, Kafka contributors are pushing a diskless-topics model that rethinks durability and replication in cloud environments, and Google officially closed Wiz acquisition in one of the biggest cloud-security moves. Plus: AWS is sunsetting Copilot CLI, Kubernetes launches an AI Gateway Working Group.

Links

McKinsey statement on Lilli

https://www.mckinsey.com/about-us/media/statement-on-strengthening-safeguards-within-the-lilli-tool

Kafka diskless topics proposal

https://cwiki.apache.org/confluence/display/KAFKA/The%2BPath%2BForward%2Bfor%2BSaving%2BCross-AZ%2BReplication%2BCosts%2BKIPs

Google completes acquisition of Wiz

https://blog.google/innovation-and-ai/infrastructure-and-cloud/google-cloud/wiz-acquisition/

AWS Copilot CLI end-of-support

https://aws.amazon.com/blogs/containers/announcing-the-end-of-support-for-the-aws-copilot-cli/

Kubernetes AI Gateway Working Group

https://kubernetes.io/blog/2026/03/09/announcing-ai-gateway-wg/

Amazon Bedrock observability for first-token latency and quota consumption

https://aws.amazon.com/about-aws/whats-new/2026/03/amazon-bedrock-observability-ttft-quota/

Cloudflare JSON responses and RFC 9457 support for 1xxx errors

https://developers.cloudflare.com/changelog/post/2026-03-11-json-rfc9457-responses-for-1xxx-errors/

Amazon S3 source-region information in server access logs

https://aws.amazon.com/about-aws/whats-new/2026/02/amazon-s3-source-region-information/

AWS Config adds 30 new resource types

https://aws.amazon.com/about-aws/whats-new/2026/03/aws-config-new-resource-types/

Amazon Bedrock AgentCore Runtime stateful MCP server features

https://aws.amazon.com/about-aws/whats-new/2026/03/amazon-bedrock-agentcore-runtime-stateful-mcp/

More episodes and show notes at

https://shipitweekly.fm

On Call Briefs at

https://oncallbrief.com

This week on Ship It Weekly, Brian covers five “AI meets reality” stories that every DevOps, SRE, security, and platform team can learn from.

Block’s AI layoff story is getting messier as follow-up reporting pushes back on the original framing, Meta bought Moltbook and brought more attention to the trust and security problems already showing up around AI-agent platforms, and Atlassian cut about 10% of its workforce while saying AI is changing the skills and roles it needs. Plus: GitHub gives one of the more honest outage breakdowns we’ve seen lately, Anthropic and Mozilla show a more grounded AI use case with Claude finding real Firefox bugs, and there’s a quick lightning round on Bedrock AgentCore policy, Dependabot for pre-commit hooks, and Cloudflare’s latest threat report.

Links

Block layoffs follow-up

https://www.theguardian.com/technology/2026/mar/08/block-ai-layoffs-jack-dorsey

Meta acquires Moltbook

https://www.theguardian.com/technology/2026/mar/10/meta-acquires-moltbook-ai-agent-social-network

Wiz on Moltbook exposure

https://www.wiz.io/blog/exposed-moltbook-database-reveals-millions-of-api-keys

Atlassian team update

https://www.atlassian.com/blog/announcements/atlassian-team-update-march-2026

GitHub availability issues write-up

https://github.blog/news-insights/company-news/addressing-githubs-recent-availability-issues-2/

Anthropic + Mozilla Firefox security

https://www.anthropic.com/news/mozilla-firefox-security

Anthropic labor market report

https://www.anthropic.com/research/labor-market-impacts

AWS Bedrock AgentCore Policy GA

https://aws.amazon.com/about-aws/whats-new/2026/03/policy-amazon-bedrock-agentcore-generally-available/

GitHub Dependabot support for pre-commit hooks

https://github.blog/changelog/2026-03-10-dependabot-now-supports-pre-commit-hooks/

Cloudflare 2026 Threat Report

https://blog.cloudflare.com/2026-threat-report/

More episodes and show notes at

https://shipitweekly.fm

On Call Briefs at:

https://oncallbrief.com