All skills
gemini-live-api-dev
Build real-time Gemini Live API voice, video, and transcription sessions with working Python or JavaScript streaming code.
- Part of google-gemini/gemini-skills
- 11.4k installs
- Developer Tools · Backend
When to use it
You have a mic or camera stream and need a live Gemini session with audio playback, VAD, or transcription wired up.
For example
You say
Using gemini-live-api-dev, write a Python script that streams my microphone to gemini-3.8-live and plays the spoken reply, with input and output transcripts printed.
You get
- Python streaming script
- Audio playback handling
- Transcript printing
How it works
- Extended thinking mode needs non-blocking function calls and interaction_status tracking.
- Sessions need context compression or resumption to outlast the ~10 minute connection limit.
Before installing
Requires google-genai (Python) or @google/genai (JS) version 2.3.0 or newer; legacy SDKs are not supported.
More Backend skills
See all- lark-shared497kHandles Lark CLI login, identity, permissions and authorization status.
- lark-event489kStreams Lark real-time events like messages, approvals and meetings for bots and subscribers.
- golang-project-layout42.8kSets up Go project folders, module names and workspaces following standard conventions.
- golang-observability42.2kInstruments Go services with structured logs, Prometheus metrics, OpenTelemetry tracing, and dashboards.
Also in this pack
- gemini-api-dev27.9kWrites Python and TypeScript code that calls the Gemini API for chat, multimodal input, media generation, function calling and structured output.
- gemini-omni-flash-api2.5kGenerates and edits video with Gemini Omni Flash, including text-to-video, frame transitions, extensions and ffmpeg preprocessing.
- gemini-api-cli475Calls the Gemini API from your shell to generate media, analyze files and YouTube videos, transcribe audio, and run research.