Connect with the Product Hunt community

DungeonQ's goal is to give security teams and authorized defender agents time and space to respond while a designated suspicious session works in a persistent synthetic world.

v0.15.0 adds a reproducible console showing the next step: what happens after an alert?

1. A real MCP client enters the workspace, checks conflicting claims and saves a note that survives restart. Wrong Tickets remain useful only inside their issuing world. A claimed "source restored" message does not change the route. Follow-up work is bounded and offers a stop; participant text cannot authorize operator changes.

2. A separate local receiver acknowledges an alert and rotates a temporary connector credential. Readbacks check that old authority and old-ticket use are denied, authorized new reads work, and those results survive restart.

This week I went through the sign-up flow of ten well-known products, imagining an AI assistant doing it for me. It could read what each product does and what it costs. On almost every one it would stop at the email confirmation or a CAPTCHA. I had never thought of a sign-up form as something software would fill in.

I see three ways to handle it, and I keep going back and forth:

1. Keep sign-up human-only. Safer, and nobody has to think about bots with accounts.

2. Let the agent do everything except one approval step, where the person confirms the email and accepts the terms.

3. Give agents their own path, like an API key or OAuth login, separate from the human form.

Hey Product Hunt!
I m Oskar, the creator of TravelerMap.
I started building it a few years ago with my brother Kacper. He helped bring the original version to life, but eventually had to focus on his other project, Veloplanner. TravelerMap then sat untouched for quite a while.
Now I m back with my business partner Lukas, and we re working on it every day.
TravelerMap is a 100% free interactive map for discovering national parks worldwide, plus other parks across the US, Canada, the UK, and Poland.
Right now, we re focusing on the US. We ve added campground and parking locations, and we re improving park descriptions with practical information for visitors.
Our biggest priority is the travel planner. We re working hard to have it ready on launch day, so you can choose parks, discover nearby stops, and save your itinerary. Alongside that, we re adding more locations, developing new features, and gradually expanding functionality worldwide.
I d love to hear what you think. What s missing, what feels confusing, and what would make planning your next park trip easier?
Thanks for checking out TravelerMap!

I maintain Termexo, a local-first Windows workbench for coding agents. This is an AI-assisted update, checked against the release notes.

One design choice in 0.10.14 is to stop treating the user opened the window as the user handled the message. Approval requests, input requests and completed tasks now keep the Windows taskbar reminder active until the corresponding message is acknowledged or cleared.

The same principle applies to setup: an install or update result remains available after closing and reopening Settings. Version, command and log details are available under advanced options. If the available version cannot be queried, the UI offers retry and network settings instead of claiming an update is available.

The setup panel covers Claude Code, Codex CLI, OpenCode, Pi Agent, Grok Build and Antigravity. Pi is new in this release, including native session restore, task execution, status events and local MCP integration. Thanks to wuyh410 for PR #33.

Hi everyone,

Vitals 1.2 is out. Most of it came from what you asked for here and by email after the launch.

It's funny watching teams panic over model safety guards while their container network policies are wide open. If an agent manages to probe internal services or reach outside connections, it s not because the model got "too smart" it s because the isolation layer was poorly built from day one.

Prompt tuning won't save a leaky environment. How does your team usually test your container boundaries before shipping?

I ve been asking sales teams this on demo calls lately, and the answers are always unintentionally hilarious.

The pricing breakdown? Trapped in a spreadsheet only one person knows how to read.
The real reason we lost that massive deal last spring? Locked in someone s head.
Objection handling? Buried in a Notion doc that hasn't seen daylight since 2024.
And the best advice for week one? "Just shadow our top rep and take notes."

So new hires spend weeks piecing puzzle context together before they actually feel ready to talk to buyers. And whenever they hit a wall, a senior rep gets pulled in to bail them out.

If you ve ever trained a sales rep (or been the new hire):

I launched NotchMind, a Mac app that turns the MacBook notch into a small hub, as a solo developer with under 500 followers on X. It finished #16 for the day with 76 points, and I'm at about 20 sales so far.

The traffic surprised me:

• X, where I post the most: the tracked link brought 1 vote.

• YouTube: my demo video has about 4.6K views, and around 73% of them came from YouTube recommending it for free. I only spent about $11 on a promotion.

• Reddit, Hacker News and Indie Hackers all stopped me because my accounts were new. One flagged me as spam, one flagged my post, and one wouldn't let me post.

I'm trying to work out where to put my time next, so two specific questions for makers with a small audience:

Curious how other builders handle the everyday S3 stuff: uploading assets, checking what s in a bucket, invalidating CloudFront after a deploy.

For me it was always a mix of the AWS console (slow for small tasks) and the CLI (fast, but I never remember the flags). Most GUI clients I tried felt dated or were closed source, which made me uneasy about handing them AWS credentials.

That frustration is why I built Hoistbay, a free, open-source desktop app that treats S3 like a regular folder and keeps CloudFront one click away. It s live on Product Hunt today if you want to take a look: Hoistbay
But I m genuinely more interested in your workflow. What do you use, and what s the one S3 task that still annoys you?

https://www.producthunt.com/prod...

Last week I asked here where you go to find out why a number moved. The answers were almost all the same: the CRM, Slack, the support desk, Linear, a doc somewhere. Anywhere except the dashboard.

That's what we launched today.

With MCP Connectors, you connect the tools your team already uses, and Genie, our AI Analyst, uses them to answer your questions. Ask why signups dropped 18%, and it checks Linear, finds the signup form that shipped Tuesday and breaks on Safari, and ties it to the drop. You get the number and the reason in one answer.

After months of building, tweaking, and probably spending far too long getting a little robot to behave, Marv is officially live.

Marv is a little Windows companion that lives around your cursor, helps you understand what's on your screen, answers questions out loud, and can draw arrows, circles and annotations to guide you through things.

Think of it as your cursor's new plus-one.

I'm building Marv to make getting help on your PC feel more natural, without constantly switching windows or breaking your flow.

Yesterday, OpenBot launched on Product Hunt and finished as the #2 Product of the Day with 350 upvotes.
Honestly, we didn t know what to expect. Seeing so many people support OpenBot, test the product, ask questions, and share honest feedback meant a lot to our entire team.
Thank you to everyone who:
- upvoted and commented,
- shared our launch,
- downloaded and tested OpenBot,
- reported bugs,
- or told us what we should build next.

The ranking is exciting, but the most valuable part was seeing that other people share our vision: AI should move beyond isolated chats toward teams of specialized agents that can work together.
We re now going through every comment and piece of feedback from the launch. Some of it has already influenced what we want to improve next.
This launch was not the finish line. It was the beginning of the next stage for OpenBot.
Thank you again for making this launch special. We re just getting started.
- Kamil and the OpenBot team

I was looking at @Higgsfield and other AI video tools recently.

You can now create an ad without organizing a real shoot. Higgsfield also has tools specifically built to generate product ads from existing assets. A normal shoot means finding a location, people, equipment, then filming and editing everything. Now you can test different versions before doing any of that.

I don't think real shoots are going away, but I can see brands doing a lot less of them.

When would you still choose a real shoot over AI?

The original idea behind GenPage was simple: if you ve taken the time to write a relevant sales email, the page someone clicks through to find your offer should feel relevant too.

So we made it easier to create a dedicated page for each prospect. Something that helps them see how your product or service fits their business.

Working with customers, we saw the same problem across their campaigns.

Someone clicks an ad for CRM for real estate agents and lands on a homepage that talks to everyone. The ad did its job. Now the visitor has to figure out the connection themselves.

Cloud Releases - Compare can now show you what the other clouds offer for the services you use, and what has changed in each of them lately, all on one page.

Try Compare: https://www.cloudreleases.io/com...

MePersona brings all your social media conversations into one place, but it goes beyond just managing your inbox.

As a creator, keeping up with comments and DMs across multiple platforms can be challenging.

With MePersona, you can:

• Manage comments and DMs from Instagram, YouTube, TikTok, X, and more in one inbox.

• Get reply suggestions that match your communication style.

• Keep track of past conversations with your fans.

• Review and approve replies before sending them.

• Connect with English- and Spanish-speaking audiences.

When my programming agent makes an error and I point out the wrong part, the corrections don't just stay in the chat but are stored as files, so the next session doesn't trip over the same thing again. I spent quite a while figuring that out actually.

In this way, I have 100 incident notes which the agent has written, and it also keeps memory notes in the project, with one fact for each file. A one-line index of those notes is loaded each time a session starts. Should the same mistake occur another time, it won't merely be a memory but a hook. That hook runs on every tool call and every reply, whether the agent remembers it or not. At the moment I have 63 hooks.

Search Central states plainly that AI Overviews need no special markup, no AI text files, no schema. The one independent study of what users actually do is 18 months old and says a citation is worth a 1% click. Everything fresher is measured by the people selling the service.

It's Q4 planning season, which means a lot of small teams are about to write GEO on a budget line. Before you do, I'd spend twenty minutes on two documents, because they disagree with the pitch in a specific way.

The first is Google's own page on AI features and your website, in Search Central. It says there are no additional requirements to appear in AI Overviews or AI Mode, and no special optimisations necessary. Further down it's blunter: you don't need to create new machine-readable files, AI text files, or markup, and there's no special schema.org structured data to add. The whole page is a restatement of ordinary SEO fundamentals - be crawlable, be indexable, be useful, have your important content in text.

That's the platform saying, in its own documentation, that the product being sold to you does not exist as a distinct technical practice. You can decide Google is being coy. I'd just note that this is the one party in the conversation with nothing to sell you, and that it's the version I'd have to defend if I were wrong.

If a model can escape, the sandbox was never actually a sandbox. Network policies were too loose, shared services created side channels, privilege boundaries were theater. No amount of prompt engineering fixes a misconfigured egress rule

Right now, to use languages like Greek, Hebrew or Swedish in Tipword, you have to set up your own Google Translate API key. Most people hate that step.

I m thinking of running it on my side instead. You d just turn it on and it works, no key and no Google account. It d cost a little extra since I pay Google for it. On-device translation stays free like now.

Vitals 1.3 is out, and the app now works in ten languages besides English:

- German

- French

This release is about keeping things in order: tabs that sit together in groups, folders that open right in the list, and iPhone photos that arrive already sorted.

Tab groups

I've been managing a dozen narrow agent configs where each file limits tools and picks a specific model like Opus for architecture planning and Sonnet for basic coding.

But assigning heavy models to review tasks gets expensive fast and juggling too many sub-roles makes the delegation chain messy.

If you use subagents in your dev workflow, do you prefer strict tool boundaries for every single task, or just two or three general-purpose agents?

Our signup numbers look great, but session recordings revealed a harsh truth: over 40% of new users drop off during our setup wizard.

We thought our step by step tooltips were helpful but they actually act like speedbumps preventing users from seeing real value quickly.

How do you audit your onboarding to strip away setup friction while ensuring users still get properly set up?