Squad CLI v0.9.4-insider: AI-Powered Teams for GitHub Repos

Some BIG updates coming for Squad that are going to make this invaluable for you to use with your GitHub Copilot toolset

Squad just keeps getting better and better—and this release is a huge leap forward. 🚀 If you haven’t tried Squad yet, you’re genuinely missing out. Setup is easy, but what your Squad can help you accomplish is next-level. Integrated watch mode. Fleet execution. Distributed AI teammates that actually scale with how you work. What more can you ask for?!

Over the past year, we at Independence Blue Cross have deployed GitHub Copilot to our technology teams and it has become a valuable force multiplier for our development teams and has helped us deliver more for less. By accelerating boilerplate coding, surfacing patterns faster, and reducing context-switching, developers focus more time on business problem-solving and higher-value work. Thanks to Larry Reeve, Robert Pruett, Timothy O'Malley, Mark Taylor #DeveloperProductivity #GitHubCopilot #AIinEngineering #SoftwareDevelopment #DevEfficiency

Sveltos v1.8.0 is out. Three new features, handful number of bug fixes, each targeting a specific operational gap. 1. 𝗚𝗶𝘁𝗢𝗽𝘀-𝗳𝗿𝗶𝗲𝗻𝗱𝗹𝘆 𝗸𝘂𝗯𝗲𝗰𝗼𝗻𝗳𝗶𝗴 𝗿𝗲𝗻𝗲𝘄𝗮𝗹 When Sveltos rotated a cluster's kubeconfig, it wrote a new key (`re-kubeconfig`) and updated `SveltosCluster.spec`. GitOps tools like Flux or ArgoCD saw this as live-state drift from the Git source. The fix is a new field: 𝘴𝘱𝘦𝘤:  𝘵𝘰𝘬𝘦𝘯𝘙𝘦𝘲𝘶𝘦𝘴𝘵𝘙𝘦𝘯𝘦𝘸𝘢𝘭𝘖𝘱𝘵𝘪𝘰𝘯:  𝘬𝘶𝘣𝘦𝘤𝘰𝘯𝘧𝘪𝘨𝘒𝘦𝘺𝘕𝘢𝘮𝘦: 𝘬𝘶𝘣𝘦𝘤𝘰𝘯𝘧𝘪𝘨 # 𝘴𝘢𝘮𝘦 Set 𝘬𝘶𝘣𝘦𝘤𝘰𝘯𝘧𝘪𝘨𝘒𝘦𝘺𝘕𝘢𝘮𝘦 to the existing key name and Sveltos overwrites the Secret in-place, skipping the spec update. Live state and Git stay aligned 2. 𝗦𝗸𝗶𝗽 𝗻𝗮𝗺𝗲𝘀𝗽𝗮𝗰𝗲 𝗰𝗿𝗲𝗮𝘁𝗶𝗼𝗻 𝗳𝗼𝗿 𝗿𝗲𝘀𝘁𝗿𝗶𝗰𝘁𝗲𝗱 𝗥𝗕𝗔𝗖 In multi-tenant clusters where Sveltos operates without cluster-wide Namespace permissions, the pre-flight namespace check would fail even when the namespace already existed. 𝘚𝘬𝘪𝘱𝘕𝘢𝘮𝘦𝘴𝘱𝘢𝘤𝘦𝘊𝘳𝘦𝘢𝘵𝘪𝘰𝘯: 𝘵𝘳𝘶𝘦 on 𝘗𝘰𝘭𝘪𝘤𝘺𝘙𝘦𝘧s and 𝘒𝘶𝘴𝘵𝘰𝘮𝘪𝘻𝘢𝘵𝘪𝘰𝘯𝘙𝘦𝘧𝘴 bypasses that check and deploys directly into pre-provisioned namespaces. 3. 𝗡𝗮𝘁𝗶𝘃𝗲 𝗛𝗲𝗹𝗺 𝘁𝗲𝘀𝘁 𝗲𝘅𝗲𝗰𝘂𝘁𝗶𝗼𝗻 Helm charts ship with test hooks that authors write precisely to validate a deployment. With 𝘳𝘶𝘯𝘛𝘦𝘴𝘵𝘴: 𝘵𝘳𝘶𝘦 in the chart spec, Sveltos runs those hooks automatically after each successful install or upgrade. Failures surface as deployment failures in ClusterSummary status. 𝗕𝘂𝗴 𝗳𝗶𝘅𝗲𝘀 𝘄𝗼𝗿𝘁𝗵 𝗻𝗼𝘁𝗶𝗻𝗴 Modifying Helm patches no longer goes undetected when chart version and values are unchanged. Sveltos now tracks a hash of the patch set. Failure messages on Helm releases are no longer overwritten before persisting. And manually deleting a ClusterSummary while a cluster is mid-update no longer causes a reconciliation deadlock. Full changelog: https://lnkd.in/d65DAXuA #Kubernetes #MultiCluster #GitOps #PlatformEngineering #Sveltos

Today I had a very bad experience with 𝐆𝐢𝐭𝐇𝐮𝐛 𝐂𝐨𝐩𝐢𝐥𝐨𝐭 𝐏𝐫𝐨+ during active development. I hit a message saying: “𝐒𝐨𝐫𝐫𝐲, 𝐲𝐨𝐮'𝐯𝐞 𝐞𝐱𝐜𝐞𝐞𝐝𝐞𝐝 𝐲𝐨𝐮𝐫 𝐰𝐞𝐞𝐤𝐥𝐲 𝐫𝐚𝐭𝐞 𝐥𝐢𝐦𝐢𝐭. Try again in 3 days.” 😁 The issue is, I still have 500+ premium requests remaining, but they are blocked due to a separate weekly cap. From a developer experience point of view, this is frustrating and unpredictable. Usage limits should ideally be aligned with available requests, not hidden weekly throttling. It broke my workflow momentum at a critical point in development. Just sharing a real-world experience from today. #github #copilot #claude

One week. One hackathon. And a very cool demo video 😂 I just shipped Envoy Watch — a GitHub App that automatically creates isolated preview environments for every pull request, then destroys them when the PR closes. Here's what that means practically: You open a PR. A bot comment appears on it within seconds. 3-7 minutes later, that comment updates with a live URL — a fully deployed version of your app, running your exact branch, in its own isolated container. You merge the PR. The environment is gone. No cleanup. No lingering infra. No cost. I built this on top of Locus's BuildWithLocus API — a platform that lets you deploy containers programmatically with a single API call. No Dockerfiles, no cloud console, no DevOps. I'd never used it before this week. The interesting engineering challenge wasn't the deployment part. It was the orchestration. GitHub doesn't talk to Locus. Locus doesn't talk to GitHub. My Next.js app sits in the middle — receiving webhook events from one side, making API calls to the other, storing the mapping in Neon Postgres so it knows which environment to destroy when a PR closes. Three moving parts. One flow. Fully autonomous. One honest note: for the hackathon, all environments deploy from my own Locus wallet. In production, each user would connect their own API key so deployments bill against their account, not mine. Scoped intentionally, the architecture supports it cleanly. The demo video says more than I can in a post. If you're a developer tired of shared staging environments breaking your team's workflow — this is for you. #buildinpublic #softwaredevelopment #africatech #nextjs #hackathon

There’s this specific kind of magic in software engineering that still completely catches me off guard. You write some code, configure a few things, and suddenly you've essentially breathed life into a piece of metal. I decided to spend my holiday weekend chasing that exact feeling. In my last update, I mentioned I was diving into Phase 3 of my HP EliteDesk homelab build. Today, that setup is officially complete, and honestly? The results are blowing my mind. I took a machine that I had to manually baby and turned it into a fully automated, self-monitoring ecosystem. Here is what the server is doing on its own right now: 🚀 Zero-Touch Deployments: I deployed a Self-Hosted GitHub Actions Runner directly on the hardware. Now, when I push code from my laptop, the server securely authenticates using temporary tokens, pulls the fresh code, and rebuilds the Docker containers completely in the background. No SSH-ing in. No manual commands. It just... works. 📲 A Server That Texts Me: I set up Uptime Kuma to keep a constant heartbeat on my applications and wired it into a custom Telegram bot. If my PostgreSQL database even blinks, the server literally sends a text message straight to my phone in real-time. Sitting back and watching a physical machine across the room deploy code on its own while the green checkmarks roll in is an incredible feeling. It reminds me exactly why I fell in love with tech. It’s a playground that never stops evolving, and there’s always a new layer of automation that completely shifts your perspective on what is possible. And speaking of the code it's deploying... TaskZilla is coming to you soon. 👀 What was the last technical concept or project that genuinely surprised you or reignited your passion for tech? Let's chat in the comments! 👇 #SoftwareEngineering #Homelab #DevOps #GitHubActions #Docker #BuildingInPublic #TechPassion

Well, it's live! https://biomelab.dev/ I've building this project to make my workday more prolific. If you are into #agentic development workflows, using git worktrees for parallel task, you probably want to take a look at this project, which helps me in monitoring the ongoing work. Spoiler alert: it uses Docker, Inc Coding Agent #Sandboxes under the hood for adding safety to your workflows 😉

GitHub Copilot is a pair programmer that suggests code snippets and full functions in real time inside your editor. It reads the surrounding code and comments to autocomplete patterns, draft unit tests, scaffold endpoints, and handle repetitive glue work. Best for developers who want to move faster and cut boilerplate without breaking flow. Use it to spike features, explore unfamiliar APIs, and standardize routine code. Guide it with clear function names and comments, review suggestions like any pull request, and keep security checks in place for critical paths. #GitHubCopilot #PairProgramming #DevTools

devpath-idp | Phase 6 update — Backstage self-service provisioning — I said I'd share more on the scaffolder issue when I had a cleaner picture. Here it is. After going back and forth on it, I realised I was starting to loop — re-proving things I'd already proven. So I stopped and took stock of what was actually confirmed. Here's where Phase 6 actually landed: ✅ The template works ✅ Backstage creates the GitHub repo successfully ✅ The GitHub token is valid — manual push with the same token works perfectly ✅ Repo creation and GitHub authentication are not the problem The one thing that doesn't work: Backstage's internal push step times out in local development. The repo gets created, but the scaffolded files don't make it in. That's a specific, isolated limitation. Not a failure of the whole phase. In a real production setup, this timeout behaviour is less likely to be an issue — local dev mode introduces constraints that don't exist in a properly deployed environment. But documenting it honestly is part of the work. So the Phase 6 conclusion is: Backstage self-service provisioning partially works. Repo creation is proven. The push step has a known local dev timeout. The limitation is understood and isolated. That's good enough to move forward. There's something worth saying about knowing when to stop debugging. Not every issue needs to be fully resolved before you move on. Sometimes the right call is to understand the problem clearly, document it honestly, and keep building. Moving on to the next phase. #Backstage #PlatformEngineering #DevOps #InternalDeveloperPlatform #GitHub #CloudEngineering #SoftwareEngineering