Find your next open-source project

Search by name, purpose, or keyword — or paste a GitHub repository URL
All skills

gemini-live-api

Scaffold a working Gemini Live API WebSocket client with session resumption, auth, and a demo UI.

  • Part of google/skills
  • 6.7k installs
  • AI Engineering · Agent frameworks

When to use it

You have a Gemini Live API project or a failing live session and need a working client, resumption, or barge-in fixes.

For example

You say

Use the gemini-live-api skill to scaffold a Python Live API client and demo UI for project my-proj in us-central1 with gemini-3.8-live, including mic audio streaming and transcription.

You get
  • Live API client class
  • SessionManager with resumption
  • Demo frontend and backend

How it works

  • Generates an isolated Python venv, setup.sh, and pinned requirements.txt inside your output folder.
  • Handles interrupt and transcription finished signals so stale audio and text don't bleed into the next turn.
  • Includes a troubleshooting path for silent turns, voice drift, tool-call loops, and WebSocket 1008/1011 drops.

Before installing

Needs a Google Cloud project with billing, the aiplatform.googleapis.com API enabled, and Application Default Credentials on the host.