Co-founder @MetalBear | building mirrord → faster, cheaper Kubernetes dev | OSS, DevEx, remocal workflows
Tel Aviv / Remote
Joined April 2022
- Tweets207
- Following140
- Followers113
- Likes965
Eyal Bukchin retweeted
We wrote about how our engineering team uses AI back in January. Nine months is a long time in this space, so @EBukchin, our CTO, revisited it with what's actually changed since.
Read here to find out which tools we're using now, and where the struggles have shifted to: metalbear.com/blog/engineeri…
k8s devs: squash the loop. Your got your agents taking coffee breaks while they wait for CI.
dear young farts out there
who have grown up with Docker/Kubernetes and the compile/restart/deploy phases of CI/CD. Did you know that the entire loop can be squashed but hasn't been because of the VC [1] industrial complex (tm)?
please stop arguging that golang/existing prior will win because it won't. nearly all programming languages are roughly the same idea but in a different bikeshed paint edition. what comes next will be an old [2] idea, [3] applied differently.
[1] lol chuckling here but somewhat directionally true over time.
[2] actors/repair the image (don't rebuild it)
[3] designed for llms as programmers, not humans
Good news for anyone who wants to make their agent evals faster, cheaper, and more accurate by running them against real dependencies!
The recording from our webinar with @dorco8 on how they do evals at @mondaydotcom is now available here: metalbear.com/events/agent-e…
And if you want our help with implementation, set up some time with us here: meetings-eu1.hubspot.com/mir…
Eyal Bukchin retweeted
@deedydas suggest solving it with old ways - more compute, more efficiency, more cache. Beyond amount of CI needed, there's almost no value in UT when using AI while tremendous value lies in E2E.
None of the listed companies help you run E2E better/faster.
The biggest cost of E2E is spinning up a real-environment.
our (@metalbear) approach with mirrord offers something different - give each test/ai/engineer a slice of a real environment, isolating flows allowing infinite scale while almost zero cost (beyond compute needed for the tests). Use it in CI, in CD and in the cloud agents it selves to have everyone iterating against same baseline.
for example - metalbear.com/mirrord/case-s…
or @fagamericano who writes a lot about it (also mirrord user :D)
As cost of coding has plummeted, long builds (CI) has become a bottleneck for many software cos. Three of the biggest startups solving this:
1. Blacksmith (@useblacksmith)
2. Namespace (@namespacelabs)
3. Depot (@depotdev)
New acronym by @nicolefv and DX just dropped! CAFE(S) defines attributes by which to measure your agentic context.
Right on the money as usual, but I had thoughts. metalbear.com/blog/cafes-age…
Eyal Bukchin retweeted
Take a read 👇
This week @dorco8 and @EBukchin's webinar left such an impression on me that I ended up writing a blog about :P
If you missed the webinar (or prefer text over video) and want to learn how @mondaydotcom runs its agent evals read the blog here: metalbear.com/blog/monday-ag…
Today I finally join my fellow CTOs with a post about @AnthropicAI's AI-Native SDLC.
My take: if you don't give your 100s of agents access to your staging env and let them go nuts, your SDLC will stay as slow as it used to be, or wreak havoc on prod.
metalbear.com/blog/ai-native…
Eyal Bukchin retweeted
Yesterday we launched mirrord Chaos Testing, which lets you break connections to dependencies in your cluster and see how your code actually responds.
@EBukchin put together a detailed step-by-step guide to getting started. If you've been wanting to try chaos testing but weren't sure where to begin, check it out here: metalbear.com/blog/chaos-tes…
Eyal Bukchin retweeted
And that’s a wrap for #KubeCon Japan 🇯🇵
Thanks to everyone who came to say hi and meet us :D
If you're looking for something to do while Claude is skedaddling or whatever, we're doing a Minesweeper thing today: minesweeper.metalbear.com/
Eyal Bukchin retweeted
In case you missed it, we just launched mirrord for Windows! To celebrate this launch, we decided to bring back everyone’s favorite retro Windows game: #Minesweeper!
Head over to minesweeper.metalbear.com/ and compete for the highest score. And if you manage to top the leaderboard, share your score in our community Slack.
The top score at the end of the week wins mirrord swag!
More details in the thread 👇
Eyal Bukchin retweeted
Launch Week Drop 2/5: mirrord for Windows 🪟
mirrord now runs natively on #Windows. No WSL, no workarounds.
If you develop for #Kubernetes on a Windows machine, you can now run your code locally while it's connected to your cluster, same as your teammates on macOS and Linux have been doing all along with mirrord!
Get started here 👇
metalbear.com/launch-week/
Only ~4 more hours until @aviramyh and @RinkiyaKeDad ‘s webinar, including a live demo of a Cursor agent iterating and self-correcting against k8s. Let’s hope it works! Sign up here: metalbear.com/events/self-co…
What's mirrord good at? Verifying code in real-world conditions, fast.
When is this useful? All the time, but especially when production is on fire, you have a proposed fix, and you want to be sure it'll actually put out the fire rather than pour gasoline on it.
AI-SREs are getting good at proposing fixes when alerts fire. The missing piece is knowing whether the fix actually works before it reaches production.
Right now there's no good answer. You have to deploy to staging to know if it works, which is slow. In our latest blog Eyal Bukchin wrote about how to fix this problem: automatically verify every suggestion against your real cluster (without deploying) before it reaches a human using mirrord.
This way:
❌ Fixes that don't pass don't get reviewed.
✅ Fixes that do come with a verdict already attached.
This approach works regardless of the tools you're using (@resolveai, @datadoghq Bits, HolmesGPT by @RobustaDev, or some in-house solution!)
Link to read the blog is in the thread
Eyal Bukchin retweeted
We'll be at @awscloud Summit Sydney next week on 13–14 May!
Working with AI coding tools? Kubernetes? Both? Come say hi 👋 and we’ll show you how mirrord helps Kubernetes teams fully leverage the productivity benefits of AI-driven development!
Eyal Bukchin retweeted
Yesterday we hosted Ryota Sawada, the Release Lead for #Kubernetes 1.36, for a webinar!
In case you missed it, you can now check out the recording on YouTube. It was a really interesting discussion where Ryota talked about:
- New and exciting features in Kubernetes 1.36
- Where he sees Kubernetes as a project heading towards in future releases
- His experience being part of the release team since Kubernetes 1.32
- How the release team works and why you should consider applying
Watch here: lnkd.in/gE9j8C3C
And if you'd like to read about the latest features in 1.36, read the blog: metalbear.com/blog/kubernete…
Outside a bunker somewhere, the MetalBear Windows team are seeing sunlight for the first time in 6 months
Windows support for the mirrord VS Code extension is coming, and early access is open now! 🪟
To try it, search for "mirrord" in the VS Code (or Cursor, or other VS Code-based IDEs) extension marketplace and download it. Then hop on our Slack and tell us about your experience!
mirrord community Slack: metalbear.com/slack
