I build ToolKnit, a browser toolbox, and have been thinking about a small design problem: the same semi-transparent color can look very different against a light or dark background. Copying a six-digit HEX value also drops alpha.

Our image color picker supports alpha-capable values and palette exports: https://toolknit.com/tools/color...

When sharing a palette, I find it useful to document both the color and the intended background, then check the resulting combination for contrast. A pleasant palette on its own does not establish accessibility, and screen colors are not a print proof.

For those who design or build color tools: what is the clearest way you have seen to show that a selected color is translucent? A checkerboard, a paired light/dark preview, or something else?

Trade Calculator puts a trade plan in a compact macOS panel: entry, stop, target and estimated profit or loss. It works with the numbers you enter; it does not connect to market feeds or trading accounts. This is a product preview.

Here is one made-up long stock trade in Trade Calculator: entry 30, default stop 3%, 100 shares, and risk:reward 1:3. Stop: 29.10. Target: 32.70. Estimated loss: 90. Estimated profit: 270. Estimates exclude fees, taxes and slippage; execution prices and returns are not guaranteed.

The default can be changed or turned off; a manual stop takes priority. How would you set the stop: a price or a percentage?

This preview copy and its explanatory graphics were prepared with AI assistance. The app screenshot comes from a native UI check using illustrative inputs.

I've been building something called Piythia, and I wanted to share the vision behind it.

I've always been fascinated by one question:

How can one person multiply their ability to create opportunities without having to do everything themselves?

We have AI models that can reason, financial markets filled with opportunities, and increasingly accessible brokerage APIs. Yet most people still have to research investments, test strategies, manage risk, and execute decisions themselves, often across multiple disconnected platforms.

Hi, I m Jeppe, founder of InReality!

I spent about 7 years researching misinformation and trying to distinguish fake content from real, including building AI detectors. But with the rise of deepfakes and generative AI, I doubt that looking for fakes is ever really going to solve the problem

Hi Product Hunt! We build mobile apps ourselves and know that getting discovered after launch can be as challenging as building the product.

We created TheWorldOf as a visual discovery platform for mobile apps. Developers choose a spot on a world map to showcase their app, and visitors can explore categories and open the relevant app-store page. The selected location represents a listing choice not a claim about where the app is popular or most used.

The site is available at https://theworldof.app. We would love feedback from both developers and app explorers: what would make a discovery platform genuinely useful to you?

We are preparing our Product Hunt launch and appreciate any thoughtful suggestions as we keep improving the experience.

I m on the team that builds AI Agent Clean, a Mac app for reviewing storage used by AI tools. In 1.0.9, we clarified the notice for unreadable locations and made it easier to see the previous scan results when switching views.

Say Deep Scan can read your logs but hits a folder it can t open. When you come back to the results, the sizes for the files it read are still useful. The missing folder s size is unknown. If the page shows a total without that context, it s easy to assume the scan covered everything it tried to read.

I've been building TASKONO, a collection of free tools for PDFs, images, writing, and work hours.

One thing that surprised me: a PDF can look perfectly readable but contain zero extractable words.

So I built an OCR tool that lets you recognize scanned pages, review mistakes, and export a searchable PDF. All in your browser, with no account needed.

A little game, because it explains what we built better than any pitch.

Picture a typical freshly shipped SaaS: Next.js front end, Supabase back end, a few API routes. You run an external scan on it and recon flags these 6 candidates. When the validation step finishes, exactly one of them is PROVEN. The other five end up refuted, inconclusive, not tested or false positive.

I m building eStockes, a stock-research workspace, and I d appreciate help testing the public dashboard: https://estockes.com/dashboard

Please open it, pick a stock you know, and try the first research task you would normally do. If you re new to the market, tell me what you would expect to do first.

1. Do you have stock-market experience, or are you new to it?

2. What did you try, and were you able to complete it?