PH热榜 | 2026-05-03

Hey PH 👋 Eyal, Roy, and Nadav here - the team behind Radar. We also build Skyhook, YC W23.

We've wanted a better Kubernetes UI for a long time. kubectl is powerful, but day-to-day cluster work still ends up split across terminals, dashboards, Helm, Argo/Flux, cloud consoles, and log tools.

The existing options all have tradeoffs. Lens lost the OSS trust that made people love it. FreeLens is a welcome fork, but still carries the same heavy Electron desktop model. Headlamp is useful, but shallow once you want deeper operations - Helm, GitOps, traffic, audits. k9s is excellent if you live in the terminal, but not everyone does. And the SaaS tools often price by node and ask for a work email before they let you look at your own cluster.

So we built the Kubernetes UI we wanted: fast, local-first, open source, and not locked behind an account. We quietly shipped it a couple of months ago. The community took it past 1.4k GitHub stars and gave us way more feedback than we expected, so we kept shipping. Today is the proper launch.

What's in it:

- Topology with real ownership chains, not force-directed spaghetti

- Live event stream across all resources using Kubernetes watches, not polling

- Helm release management with diff/rollback + native Argo CD / Flux sync

- Live traffic flows via Hubble/Cilium, Caretta, or Istio

- Cost insights via OpenCost - auto-detected per namespace, workload, and node

- Cluster audit - 31 checks across security, reliability, and efficiency

- Image filesystem viewer - read container files in the UI, no exec, no pull

- Built-in MCP server - point Cursor or Claude at your cluster

Plus first-class integrations for 20+ popular K8s tools - Argo Rollouts, Karpenter, KEDA, cert-manager, Trivy, Kyverno, Velero, Knative, and more.

Single Go binary. Apache 2.0. No account required. No usage tracking. No cloud dependency.

Site: https://radarhq.io

Repo: https://github.com/skyhook-io/radar

Discord: https://radarhq.io/community/chat

Also yes, we cared about making it beautiful. K8s tools don't have to look like punishment.

Would love your feedback - what's missing, what breaks, what we got wrong. We're here all day.

a cluster audit with 31 checks across security and reliability is a massive value-add for day-to-day ops. we usually run separate trivy or kyverno reports, so having that integrated into the primary ui workflow is a huge time-saver. does the audit allow for custom check injection? @nadav_erell

Looks very useful, sharing it with my team

nice - gave it a star! good luck!

Local-first with zero cluster-side installation is the right call. When I was scaling an engineering org from 15 to 120, the biggest friction with K8s tooling was always the chicken-and-egg problem: you need cluster access to install the tool that helps you understand the cluster. Having this run as a single binary using existing kubeconfig means any engineer can get visibility without needing platform team approval first. That alone probably saves a week of onboarding time per new infrastructure hire.

The MCP-for-AI-agents piece is the most interesting bit and the existing comments haven't poked at it. What can an agent actually do via Radar's MCP — read-only stuff (describe pods, fetch events, summarise an outage), or destructive ops (kubectl delete, rollout restart)? That boundary decides whether anyone runs this against a prod cluster.

👋 Hey Product Hunt!

I’m Sina, founder of PandaProbe.

Building AI agents is getting easier, but understanding and trusting them in production is still hard.

Once agents start calling LLMs, tools, APIs, MCPs, and sub-agents, logs aren’t enough anymore. You need to see what happened, why it failed, whether quality regressed, and how reliable the system is across full sessions.

PandaProbe is my attempt to solve this: an open-source agent engineering platform for tracing, evaluation, monitoring, and debugging AI agent applications.

The goal is simple: help developers move from “it works on my laptop” to “I understand production behavior, can measure quality, and continuously improve it.”

What PandaProbe provides

🔎 Trace — capture full agent executions as sessions, traces, and spans across LLMs, tools, agents, and custom logic.
📊 Evaluate — score traces and sessions using mission-critical, agent-specific metrics.
⏱️ Monitor — schedule recurring evaluations to automatically validate new traces and sessions in production.
📈 Analytics — track performance, cost, latency, errors, and quality trends over time.
🛠️ Open source + cloud — use the open-source core on GitHub or run PandaProbe in the cloud.

Who it’s for

🧑‍💻 AI engineers — debug agent behavior across LLMs, tools, and workflows.
🏗️ Platform teams — monitor quality, regressions, and reliability in production.
🔬 Builders experimenting with agents — understand failures and iterate faster.
🚀 Startups — add observability and evaluation before things become unmanageable.reason about.

Quick links

GitHub: https://github.com/chirpz-ai/pandaprobe

Docs: https://docs.pandaprobe.com

Cloud: https://www.pandaprobe.com/

I’ll be here all day answering questions and collecting feedback.

If you’re building agents today, what’s the hardest part to debug or evaluate?

Thanks for checking it out 🙏
— Sina

Handling state and debugging for long-running autonomous agents is usually a nightmare, so having an open-source platform to standardize that workflow is huge. I can definitely see myself using PandaProbe to self-host my agent evaluation pipeline to keep sensitive client data entirely local. I am really curious to hear if you currently support custom tracing for raw API calls instead of just the standard frameworks.

Evaluation is the hardest part of this whole space and most platforms hand-wave it. The failure mode that actually bites in production isn't crashes or schema errors. It's slow drift in subjective quality (voice, classification accuracy, output style) that only shows up when a human reads 50 outputs in a row. How does PandaProbe handle that in practice? LLM-as-judge with custom rubrics, human-in-loop on a held-out set, embedding-distance from a golden corpus, or something else? And how do you stop eval cost from outpacing inference cost when you're re-judging every trace?

Great pain to tackle, Sina. Good luck.

Really nice work. The gap between "it ran" and "I understand what happened" is enormous for agents and nobody's solved it cleanly yet. Rooting for you!

Congrats on the launch and thanks for using mcp-use :)

Where does PandaProbe sit relative to LangSmith, Langfuse, and Helicone? They all claim "agent observability" but mean very different things underneath — some are basically prompt loggers, others actually trace tool-call DAGs. Curious which problem you decided was the real one.

Hey Product Hunt!

I am Ayush, founder of Huddle01 Cloud. Virtual Machines are an essential when it comes to taking your website or applications online. In last few months, developers and non-devs have been building their applications using AI assistants. Anyone who is not familiar with tech has to spend a lot of time reading through docs to setup their Virtual Machines to get the website published. 

So we decided to fix that. 

Huddle01’s MCP server allows you to access the Virtual Machines from your AI Assistants like Claude Code, Cursor, Antigravity and any tools that support MCPs. You just need to install the server with one line of code and then your chat can handle everything else. 

With this:

• Beginners and non-developers can setup Virtual Machine instances from their AI assistant chats.

• Developers and Builders can setup their agents to auto-deploy instances for the apps that agents create.

• Check the already running instances and info about VMs without visiting the dashboard.

Huddle01 Virtual Machines come with dedicated AMD EPYC vCPUs, NVME storage and unlimited egress across global regions. 

Why Huddle01 VMs for your agents?

• MCP native: We’re the first VM provider that allow you or the agents to just write in the chat and get the infra live. 

• You don’t pay any brand tax. Just pay for what you use with no egress markup. This makes Huddle01 almost 70% cheaper than other hyper scalers. 

• Only three steps setup an then you can get VMs by chatting anytime. 

How to use: 

• Just go to get.huddle01.com and copy the code

• Paste it in your Claude Code terminal, Cursor, Antigravity or any tool. 

• Run it and you’re ready to get deploying from the chat 

If you’re building your first python project and don’t want to configure everything manually or you have a multi-agent system working that self deploys everything, Huddle01 has got you covered. 

You can leave feedback or ask questions to us in our Discord too. 

It’s time to get deploying with Huddle01 (huddle01.com). 

We’re here the whole day to answer all your questions!

Congrats on shipping!! Always exciting to see infra innovation happening.

Congrats on your launch! I'm not very knowledgeable about virtual machines or why I would need one. My current workflow to publish my vibe coded apps is Antigravity => Github => Vercel. That seems pretty straightforward and simple.

What use case are you solving that can't be done with my workflow?

I've faced this problem myself, being confused about deploying side projects or getting my agent to do it. It's been incredible to see the team ship this in such a short time.

Is there support for GPU instances as well?

Nice! But how do you ensure cost control if agents keep spinning up instances automatically? 😅

We had so much fun building VMs for Agents. Excited to see how people use it to deploy their apps!

Per-second billing only matters if spin-up is also fast — what's the cold-start time for a vanilla Ubuntu image, and does it stay flat as the image gets heavier (e.g. with a 2GB Docker layer baked in)? The economics break the moment startup is 90 seconds and I'm paying for warm-up.

Hi Product Hunt 👋

I’m Egor, co-founder of Mockin.

In the 9 months since our last launch here, the design job market got even harder.
After speaking with more than 60 designers, one pattern became impossible to ignore. Strong skills do not guarantee interviews. People get filtered out at every stage, often without understanding why.

Resume.
Portfolio.
Recruiter screen.
Hiring manager interview.
Case study.

One weak point is enough to lose momentum, and usually the only feedback you get is silence.

That is exactly why we built Mockin 2.0.

We turned Mockin into a tool that helps UX/UI and Product Designers see real hiring risks early, fix them before real interviews, and prepare across the full job search funnel based on real expectations from the market.

I’d love for you to try it and tell us what feels strongest, weakest, or still missing.

Thank you

Hey Product Hunt community! We significantly improved our app since the last launch. Appreciate your feedback on the updates and Mockin itself ♥️

Thanks to the team for a great job!

Andrey & team are awesome and always ship delightful products!!

Congrats with launch! So, now all the features to land a next job is inside?

Guys that's great for designers! Will you go to other roles?

a only 'feedback you get is silence' line is the most painful part of the 2026 job market. You can have a world-class portfolio, but if you’re failing at the recruiter screen without knowing why, you’re just spinning your wheels. Excited to see how Mockin identifies those specific hiring risks before they turn into rejections. Support on the launch, @lipkovskiy Egor

Great product

Hi, congrats for your launch. Why did you started with the UI/UX niche. Any user feed back or based on personal evidance ? I used to have a very similar idea but for other roles. UI / UX designers are the hardest role to start with I guess.

Congratulations. And happy product launch. @lipkovskiy

awesome launch, congrats!

Most career-prep tools double down on the polish layer — portfolios, mock interviews, salary scripts — and skip the actual high-leverage decision, which is which role to take in the first place. I'm an Accredited Financial Counselor and the most common thing I counsel new grads on is the "safe offer vs. the role that builds the skill stack" tradeoff — the comp gap usually closes within a year if the second role is right. Curious — are designers showing up to Mockin pre-offer to make that call, or mostly post-offer for prep?

Hey Product Hunt 👋. I'm Laramie. Nearly 20 years in tech. By day I run technical partnerships. By night I build with Claude Code and Codex on the same repo.

Here's what I kept running into: both agents write good code. Both PRs pass CI. Then they merge and break main. CI checks one branch at a time. Git merges text, not logic. Nobody checks whether the branches are compatible with each other.

With agent orchestration tools shipping weekly and PR volume exploding, this is only getting worse.

So I built Rosentic.

It checks every PR against every other active branch before merge. Deterministic. Same scan, same result, every time.

Most scans come back clean. That means you're clear to merge. When something fires, you catch it before production does.

150+ repos scanned. Real conflicts found and fixed by maintainers. One YAML file, no signup, 60-second install.

What's the messiest merge conflict you or your team has dealt with lately?

https://github.com/marketplace/actions/rosentic-cross-branch-compatibility-check

This is solving a problem that's about to hit every engineering team at scale. When I was CTO running 120 engineers, merge conflicts between humans were already one of our biggest velocity killers - now multiply that by AI agents working in parallel across the same codebase. The hard part isn't detecting the textual conflict - it's catching the semantic breaks where two agents make changes that individually pass tests but together introduce subtle regressions. Curious how you're handling that semantic layer versus pure git-level conflicts?

"Deterministic analysis" is the line that earns trust here — what's the actual technique? AST-level diff, type-graph reachability, an SMT solver, something else? Asking because "deterministic" in this space sometimes means "we wrap an LLM and average," and the value is very different depending on which one it is.

Nice work. Narrow but useful problem. Curious how you handle false positives when two PRs touch the same surface for legitimately different reasons (e.g. coordinated multi-agent refactors). That's been the trickiest part of PR-conflict detection in our orchestration work.

as an ev driver, the 'black box' feeling is real. manufacturer apps are notoriously bad at showing the actual cost-per-trip vs energy loss. being able to see energy use and charging insights directly on the dashboard instead of a phone app is a huge safety and ux win. @laurenz_hinterholzer

Quick question on data ownership: if I switch cars in 3 years, can I take trip history with me, or does it stay tied to that specific vehicle? And any plan for CSV export for mileage tax write-offs? Niche use case but it tends to be the thing that converts hobby users into paying ones.

Innovative idea – I haven't seen something like this on PH before. :)

Love the concept mate... Wish this was part of Google Maps. I'll absolutely give it a try on iOS.
1 tiny feedback on the website. I'd recommend adding a QR scanner on the bottom right of the website.

This would make it easier for folks on desktop to download the app.

Best of luck @laurenz_hinterholzer

Awesome idea! Would be great if you could add some kind of adventure mode for those who love driving but want an alternative / more scenic route for their regular routes.

Damnnnnn. I'm working on an app called Habit Doom.
I moved away from Lottie animations cause I needed to keep the app size less than 50 MB.
Will give it a try. Is there no quality loss on compression?

98% compression on Lottie JSON — is the technique mostly path/coordinate quantization, dropping redundant keyframes, or actually re-encoding to dotLottie's binary format? The answer matters because each of those has a different visual-fidelity trade-off, and "up to 98%" is the kind of number that's true on a specific shape of input.

Paste-a-link → portfolio is a great demo, but the interesting question is where it breaks. What happens with a stealth-mode startup that has no website, a portfolio entry that's only on AngelList/LinkedIn, or a company that pivoted three times? Trying to figure out whether this is genuinely "one click" or "one click, then 20 minutes of cleanup."

Where do the hundreds of random files I haven't filed away yet live then?? :) Seriously, I can definitely use this for my complete mess of a desktop. Good luck with the launch.

This looks like a messsss!!! In a good way...
Just like my mind 😅 Need something like this for my Mac.

....
Anyway, best of luck @divyesh Rooting for you 🍀

All your work eventually happens in files and folders. So instead of relying on other apps to keep track of your notes and tasks, why not turn your folders into a personal dashboard? A sticky note placed right next to your file will never be forgotten.

But it isn't just a productivity tool. You can also design and decorate aesthetic boards that track your life or turn it into the coolest launcher for games.

There's something special for everyone, whether you are a developer, gamer, student, or professional.

Looks cool, got support for firefox?

Such a cool extension. For someone that has tens of tabs open at all time for a variety of tasks, I love that I don't have to switch to an entire different browser (which means loosing mobile sync and potentially other feature) just to modernize my tab situation. Keeping tabs manually organized gets out of control too quick to be a feasible option, and the suspend feature on top of that is (unfortunately) perfect for the modern web

❤️

This is amazing. I’ve just tried this for the first time and after trying countless tab managers this is the only one that works for me. Great job!

Hey everyone, Ciprian here, the developer of Uncluttr!

This is the first time I ever make a personal product public. For a long time I struggled with having too many tabs open.
I would close all of them and in 6 hours max I would get back to 60+ tabs.

About a year ago I started building my own solution, a vertical sidebar tab manager with AI grouping, so I'd never have to manually organize my tabs again.

Why not use an existing one? I tried. Most tab managers are built to be generic. AI changes that, it learns how you work and organizes tabs around your actual workflow, not a one-size-fits-all system.

After using it daily for months, I realized this could work for others too. So I put in the extra work to make it public.

What you get for free:

• vertical sidebar

• unlimited workspaces and groups

• tab suspension to save memory

• tab search

• deduplication

• no account needed, all data stays in your browser.

• a couple of AI groupings, enough to get a taste of it's power

To celebrate the launch, I'm offering the first 6 months free on the yearly plan with promo code PH50.

I'm actively building this and your feedback directly shapes what comes next. What would your dream tab manager do that nothing on the market does today?

Nice one!

In my day to day work as an analyst, I have lots of tabs open (easily 30+ just for JIRA, plus several others for documentation and research). Needless to say sometimes I get overwhelmed. Imagine when I share my screen and I can't find some tickets...

I am looking forward to the cross-device synching. When I'm away from my desk, my phone basically becomes my laptop.

Hey Product Hunt! 👋

Blurts started as a tool I desperately needed for myself.

I’m a full-time software engineer, real estate property manager, and mom to a one-year-old daughter. My brain generates “don’t forget to do X” thoughts at the worst possible moments - while holding the baby, in a meeting, fighting a diaper change, walking between rooms, or too sleep-deprived to remember what I was doing in the first place.

And the problem is: I almost never have one clean thought at a time.

It’s usually four thoughts tangled together:

“Fix the leak, push the PR, buy diapers, and reschedule the dentist to next Tuesday.”

I tried voice memos. They captured the chaos, but they didn’t organize anything. Then I had to come back later, listen through everything, and turn the recording into actual tasks myself.

I tried notes and to-do apps too, but they expected me to stop, type clearly, choose the right place, pick the right fields, and organize everything manually. By the time I finished capturing one thing, I had usually forgotten the next one.

So my husband and I built Blurts.

Not as a shiny “AI productivity platform,” but as a simple way to get messy thoughts out of your head before they disappear.

🎙️ Speak naturally
Tap one button and say everything at once. Ramble, pause, repeat yourself, use filler words — no need to sound organized.

🧹 Blurts cleans up the word soup
It strips out the filler, separates mixed thoughts, and turns them into clean tasks.

📆 It understands dates from what you said
If you say “next Tuesday” or “tomorrow morning,” Blurts pulls that out and adds the right timing.

🔁 It works with the tools you already use
Blurts checks the structure of your existing tool and suggests fields like category, priority, status, effort, or whatever properties you already have set up.

Right now, Blurts integrates with Notion, Google Tasks, Google Calendar, Apple Reminders, and Apple Calendar.

That’s it. No complicated setup. No big dashboard to manage. Just speak, pocket your phone, and go back to what you were doing.

This is built by a husband and wife in the margins between full-time jobs, nap schedules, bedtime routines, and all the tiny life admin tasks that never seem to end.

I built Blurts because I needed something that could keep up with real life - not the calm, organized version of life, but the messy one.

If you’ve ever said “I’ll remember that later” and then absolutely did not, I’d love to hear:

Where do you usually remember tasks at the worst possible time?

And what tool do you wish was easier to use by voice?

Thanks for checking out Blurts 🧡

"Speak your chaos" is the right frame — the verbalization is the hard part, not the to-do list. I counsel clients on money and the same pattern shows up: the avoidance loop breaks the moment someone says "I have $40K in credit card debt" out loud rather than circling it silently for a week. Have you thought about a vent mode that captures the dump without converting any of it to tasks? Sometimes the tasks come later — after the unload does the actual work.

Hey everyone, really excited to finally share PostGun with you today.
This is one of the projects from my challenge of building 24 startups in 12 months, and honestly it's one I've been wanting to build for a long time.
I built this because my own content workflow was driving me crazy. For every post I wanted to publish, I had to design it in Canva, download the file, open a scheduler, reupload it, write captions for each network one by one, and somehow keep track of what was going where. By the time anything went live, I had already lost the energy I started with.
The thing is, scheduling was never my real problem. The hard part has always been building posts that actually look native to each platform. Most tools out there only solve the scheduling side, so I wanted something where the creation, the remix and the publishing all happen in the same place.
PostGun started as a simple post composer. Then I added a built-in canvas, because flipping between tabs to design killed my flow every time. After that came the remix feature, which lets you take any viral post and rebuild it into your own in a couple of clicks. That part actually changed how I think about content, because most great posts are really just smart remixes of something that already worked. Finally I connected the whole thing to TikTok, Instagram, YouTube and the rest, so you can ship across every platform from a single calendar.
The idea behind PostGun is simple: build once, remix the rest, post everywhere.
I'd love to hear what you think, what feels missing, and which platforms you'd like to see next. Every piece of feedback genuinely shapes where this goes from here.
Thanks for taking the time to check it out.

Love the concept in general but perhaps in my experience over the last few months what works on 1 platform rarely works on another. Instagram carousels that work are generally different from LinkedIn presentations that get engagement. Is there a way the same content can be rephrased and worked automatically for each platform? What do you think?

Anyway, best of luck @pasho Rooting for you 🚀

Hey Product Hunt 👋

We’re building Posting Machine to help B2B founders turn LinkedIn into a founder-led sales channel.

It helps founders draft posts from Slack, improve content with analytics, and turn engagement into outreach opportunities.

We have a promo code for Product Hunt users. Comment below or message us if you’d like to try it.

Would love to hear your feedback, questions, or anything you think we should improve.

The 3 options before committing to a message or reply is good!!!
But I want to create posts and comment on existing posts that have a problem I'm trying to solve on LinkedIn.
I'm not a slack user but I want to automate my LinkedIn engagement. Is that part of your roadmap?

Best of luck guys @ian_hsiao @hank_wu_1999

I downloaded the app.

Super easy UX but I didn't understand how to change the voices.

Love the simple onboarding.

Best of luck @caroline_zhu1 Rooting for you ✨

Thanks a million, Liviu Chita! I’m so glad you enjoyed the globe and the soundscape. All of you encouragement and Kenshi Yonezu’s music are the two pillars that support my work. Best of luck to you too! 🌍✨

Thank you so much,David! I’m incredibly moved by your kind upvoted. Your support, along with the music of Kenshi Yonezu, is what keeps me going and inspires my creativity every day. It means the world to me that you noticed the details like the sound and animations!

Thank you for the amazing feedback、Richard! I put a lot of heart into the intentionality of this project. My creative journey is fueled by supporters like you and the profound music of Kenshi Yonezu. Knowing that the globe and sound resonated with you gives me so much energy to keep building!

Such a beautiful website!!!!

Love the intentionality behind the interactions.

Loved the globe animation. It's a chefs kiss.

Even the background sound it heavy.

Best of luck mate!!! Rooting for you 🌎

Can you choose a style or background that matches a shop’s existing vibe?