Can you break this open-source real-time coordination tool in seven minutes?

#webdev #a11y #opensource #testing

I maintain ChantLive, a free and open-source tool for coordinating short, live messages at demonstrations and community events.

An organizer prepares a list and pushes one item at a time. Participants open a public link, without creating an account, and see each update in real time. The source is on GitHub.

The stack is React, TypeScript, Express, PostgreSQL, Drizzle ORM, and Socket.IO. The interesting part is not the stack, though. It is whether the participant flow survives the messy conditions of a real event.

The seven-minute break test

You can do this alone with two browsers or devices.

Open ChantLive and create a fictional event called Community Test.
Add three harmless sample messages: Welcome everyone, Together we can, and Thank you for being here.
Open the ordinary participant link in a second browser or device. No participant account is required, and you do not need to use the QR code.
Change the current message three times from the organizer view.
Refresh, briefly lock and unlock the participant device, or close and reopen the participant page.

Please do not use a real event name, personal information, confidential material, or sensitive chants.

What I want to learn

A useful result is not "looks good." I would value one concrete failure or hesitation:

Where did either role need an explanation?
Did an update arrive late, disappear, or recover badly?
What is the biggest reason you would not trust this at a real event?
What one change would make it more useful or accessible?

If browser, device, keyboard-only navigation, zoom, or assistive technology affected the result, please mention it. A short comment is enough.

Why I am asking here

Earlier outreach asked organizations to recruit two people and evaluate an unfamiliar product. That produced conversation but no completed tests. This version removes most of that work: one person, two browser windows, harmless sample data, and a focused failure-finding task.

ChantLive is MIT-licensed. There is no paid plan, endorsement request, mailing-list requirement, or partnership pitch. I am looking for evidence about what breaks and what should be improved before anyone relies on it.

Live app: https://chantlive.online/

Source and issue tracker: https://github.com/barsham/ChantLive

If you try it, please leave one specific observation below or open a GitHub issue. I will report back on what changes as a result.