Using G2 smart glasses to work on Repls despite no Replit API?

Workspace
1

Hi everyone,

I’m exploring whether there is any practical way to use Even Realities G2 smart glasses with Replit.

For context, the G2 is a pair of lightweight smart glasses with a small heads-up display, touch input, microphone input, and a phone-based SDK. Apps run mainly through the phone/companion app, while the glasses act as a display and input device.

Since Replit does not seem to have a full public API for controlling Repls, editing files, running commands, reading console output, etc., I’m wondering if there is any workaround for building a G2-based workflow.

Possible approaches I’m thinking about:

  • Using Replit’s web UI through some kind of browser automation layer

  • Using GitHub/Git as the interface instead of directly controlling Replit

  • Using SSH or shell access where available

  • Building a separate backend that talks to the Repl indirectly

  • Using Claude/OpenAI as a coding agent, then syncing generated code into Replit

The rough goal would be something like:

  • View Repl status or console output on the glasses

  • Dictate simple coding instructions

  • Trigger small edits or commands

  • Use the glasses as a lightweight coding/control interface, not a full IDE replacement

Has anyone tried something similar, or is there a known way to programmatically interact with Repls despite the lack of an official API?

Any pointers on realistic architecture, limitations, or “don’t waste your time” warnings would be appreciated.

2

Could anyone nudge me in the right direction please?
Whether this is even potentially plausible?

3

Replit is a tightly controlled system, they don’t have an API from what I know. I did this with SSH when i was using replit with my own servers but I found it was pointless if all I wanted was the automatic deploy function, I just built my own infra, scripts and managment system with claude/codex and realized I didn’t need replit for what I was doing. I’m not saying don’t use Replit i’m just saying once I started breaking Replit open and trying to cherry pick features I realized it’s just a harness and don’t really need it. For more advanced users we like the idea of working within these systems until we want to “mod them” and then we realize we should just build the engine ourselves so we can supercharge it and add all of he things we want.

1 Like
4

Hi Jeff,

Thanks for your feedback.

I was hoping for some sort of ui where I can toggle between repls to work on the most recent agent responses, but I guess without an API, you run into deadends pretty quickly.

Replit was revolutionary as a vibe-coder with zero programming knowledge.
But I AM looking to level up to the next tier of vibecoding with a more customized workflow with a dedicated multi-agent workflow while keeping costs down. Pretty big tangent but do you know where I should start to learn more info about that?