Hi everyone! I’m developing The Empty Frame, a historical detective mystery, and I’m looking for honest feedback on the English beta.

A painting has vanished. Five witnesses give conflicting accounts. You examine visual evidence, question witnesses and connect clues across two acts.

Project page: https://honzavitha.itch.io/the-empty-frame

Use the “PLAY THE FREE ENGLISH BETA” link on the page. The game runs on an external website, with no payment, download or player account required.

I’m especially interested in: • Where did you get stuck or lose interest? • Were any clues or instructions too obvious? • Did the deductions feel fair, including the evidence the game accepted? • How readable and easy to use was it on your device? • If you tried the optional witness audio, did the voices suit the characters?

Even a partial playthrough would help. Please include your device/browser, approximate playing time and an overall rating out of 10. If you have a price in mind for a polished, complete case, please include the currency—zero is a valid answer too.

You can reply here or use the in-game Feedback form, which does not require enabling gameplay tracking. Please mark spoilers in public replies.

For transparency, AI tools assisted with the story, code, artwork, synthetic voices and English adaptation. The mystery has a fixed story and solution.

Thanks for helping me find what needs improving!

Spoilers: opening evidence files below.

AI-assisted partial playtest from NextGenRunGames: desktop in the Codex in-app browser, about 3 minutes. I chose Play without tracking and read three of the four opening evidence files, then stopped before conversations or deductions because I had not tested the reasoning step. I cannot rate the complete case or suggest a price from this check.

The first two files gave me a useful timeline to reason from: Béroud finding the wall empty on 22 August establishes discovery, while the next file dates the theft to 21 August and explains that the museum being closed to visitors did not exclude staff access. That feels like a concrete, evidence-led next question without giving away a culprit. The third file also says the empty frame does not establish that the wooden panel was destroyed, which helped keep that inference bounded. The opening text was readable in my desktop view; I did not try the optional audio. I was not stuck in these files; I stopped before testing either deduction, so I cannot judge whether the final solution is fair yet.