How it works (high level): a realtime voice agent builds out Github Issues with "@claude" appended; Claude Code’s GitHub Action executes coding tasks headlessly in the cloud, creating a PR that our voice agent can then merge back in or continue iterating on.
Why I built it: I live and work from an RV. I have a ton of free-time while driving and wanted a way to prototype and write code without a keyboard.
What’s different: Entirely client-side app with bring-your-own-key support Free tier: 5 conversations/day; one-time payment unlocks unlimited conversations.
Looking for feedback on: latency, failure modes, the prompts you’d want pre-baked, and tools you'd like integrated.
Link: https://www.voicescri.pt Demo: https://www.youtube.com/watch?v=vUGgWjYQGIE
Happy to answer questions!
p.s. The first 90 customers can use the LAUNCH90 coupon for 90% off.