Persona Hits 707 Stars in Three Days by Giving Voice Agents a Face Without Touching the Microphone
xikhar/persona (created 2026-07-28, 707 stars and 59 forks in three days, MIT for the app, pushed 2026-07-31) renders an animated VRM character whose expressions sync to voice output, using native per-application audio-output listeners — PipeWire on Linux, WASAPI on Windows, Core Audio on macOS 14.2+. Its privacy stance is unusually explicit for the category and is stated as a capability boundary, not a policy: Persona does not capture the microphone, save audio, produce speech, transcribe content, or send audio over the network. It integrates with agents like Codex over MCP and exposes an HTTP loopback API, and users can import their own VRM models and VRMA animations — the bundled character assets are excluded from the MIT license and marked test-only.
Source
↳ Follow the thread