gemini-live-api-dev Skill for Claude: What It Does and How to Install It

Makes Claude build real-time voice and video apps with the Gemini Live API, using current model names and SDK rules.

Installs: 10.4K; on skills.sh
Stars: 4.4K; on GitHub
Files: 1; besides SKILL.md
Updated: Oct 2026; last repo commit

This skill from the google-gemini/gemini-skills repo gives Claude the current rules for the Gemini Live API. SKILL.md lists four current models, the Python and JavaScript SDK versions, audio formats, quick-start code for text, audio and video, and session limits.

It also covers background reasoning, live translation and live transcription. A migration guide handles legacy Live models. The folder bundles no scripts.

Good for

Developers building voice agents with the Gemini Live API
Teams adding live transcription or live translation
Engineers moving older Live models to Gemini 3.8 Live
Web developers who need ephemeral tokens for browser clients

Install gemini-live-api-dev

Claude Code

Copy this prompt.

Install the gemini-live-api-dev skill in Claude Code. Run this command:

/plugin marketplace add google-gemini/gemini-skills
/plugin install gemini-skills@gemini-skills

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Pick one route. The plugin routes install the gemini-skills plugin: 3 skills (gemini-api-dev, gemini-live-api-dev and gemini-omni-flash-api, counted from the repo tree). The skills routes install just this skill. The skill is guidance for code you write with the Gemini API, and you need a Gemini API key for that code.

From the project README. Source.

What's in the skill

When it loads

Use this skill when building real-time, bidirectional streaming applications with the Gemini Live API, or migrating legacy Live models (2.0/2.5/3.1) to Gemini 3.8 Live. Covers WebSocket-based audio/video/text streaming, voice activity detection (VAD), background reasoning (extended thinking), asynchronous function calling, session management, ephemeral tokens, live transcription, and live translation. SDKs covered - google-genai (Python), @google/genai (JavaScript/TypeScript).

Sections in SKILL.md

Overview
Models
Current Models (Use These)
SDKs
Partner Integrations
Audio Formats
Quick Start
Authentication
Connecting to the Live API
Sending Text
Sending Audio
Sending Video
Receiving Audio and Text
Background Reasoning (Extended Thinking)
Live Translation (Gemini Live Translate)
Model

Key facts

Author: Google (google-gemini org repo; the plugin manifest names Google as author, and the README says it is not an officially supported Google product)
License: Apache-2.0
Installs on skills.sh: 10,446 (Nov 2, 2026)
GitHub stars: 4,367 for the whole google-gemini/gemini-skills repo (Nov 2, 2026)
Last repo commit: Oct 30, 2026
What it includes: 2 files including SKILL.md: references/migration.md. No scripts
skills.sh audits: Gen Agent Trust Hub: pass; Socket: pass; Snyk: pass (Nov 2, 2026)
Part of: The gemini-skills plugin (version 2.2.0) in google-gemini/gemini-skills: 3 skills counted from the repo tree
Needs: A Gemini API key for your own code; google-genai 2.3.0 or later (Python) or @google/genai 2.3.0 or later (JavaScript), per SKILL.md
Works in: Antigravity, Claude Code, Cursor and OpenAI Codex, per the README. The README says it is not an officially supported Google product

What it includes

Four current models: 3.8 live, 3.8 extended thinking, 3.5 transcribe, 3.5 translate
Models
Connect, send text, audio (16 kHz PCM) or video (JPEG), and read audio and transcripts
Network · Quick start
gemini-3.8-live-extended-thinking with thinking levels and non-blocking tools
Track interaction_status (IN_PROGRESS or IDLE) instead of turn_complete alone.
Background reasoning
Speech-to-speech translation across 70+ languages with TranslationConfig
Network · Live translation
Streaming speech-to-text with interim results; smart or verbatim mode
Network · Live transcription
Text or audio per session, session length limits, ~10 min connection, 128k context
Limitations
Nine rules, from using headphones to ephemeral tokens for browser clients
Best practices
Six: LiveKit, Pipecat, Fishjam, Vision Agents, Voximplant, Firebase AI SDK
Partner integrations
Checklists for moving older Live models to Gemini 3.8 Live (references/migration.md)
Migration guide
Use the search_docs tool if present; otherwise fetch aistudio.google.com/llms.txt
Network · Documentation lookup

From the official docs

Permissions

Low risk

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

Keep the API key out of browser code
SKILL.md says to use ephemeral tokens for client-side deployments and never expose API keys in browsers. Make sure generated front-end code follows that. github.com
The plugin route adds a skill with scripts
The gemini-skills plugin also installs gemini-omni-flash-api, which bundles four Python scripts: upload_file.py, generate_video.py, inspect_video.py and prep_video.py. The single-skill routes do not add it. github.com

Security advisories

No published security advisories as of Nov 2, 2026.

Technical details

Turning on read-only	No read-only mode. Use version control.
Talks to	Gemini Live API over WebSockets, from the app Claude writes. aistudio.google.com documentation pages. Package registries for pip and npm
Sign-in and keys	SKILL.md shows an API key passed to the client in code and does not describe where to store it.
Scopes	Claude can run pip install -U google-genai or npm install @google/genai when you build a Live API app Claude can fetch Google documentation, including aistudio.google.com/llms.txt, when the search_docs tool is missing

Questions

How do I install gemini-live-api-dev?

Use the install box at the top of this page. The README documents a Claude Code plugin and a Codex plugin for the whole gemini-skills set. The box also shows commands that install only this skill.

What does the gemini-live-api-dev skill do?

It gives Claude the current rules for building real-time voice and video apps with the Gemini Live API. Claude uses it for model names, SDK calls, audio formats, session limits, translation and transcription.

Do I need a Gemini API key?

The code Claude writes needs one. The quick start builds a client with an API key, and the best practices say to use ephemeral tokens for client-side deployments and never to expose API keys in browsers. The skill itself bundles no scripts.

Is this an official Google skill?

The plugin manifest names Google as author. The repo README also says this is not an officially supported Google product.

gemini-live-api-dev or gemini-api-dev?

Use this one for streaming voice and video over WebSockets. The README describes gemini-api-dev as the skill for text generation, chat, function calling, image and speech generation and agents.

Which models does it cover?

SKILL.md lists gemini-3.8-live, gemini-3.8-live-extended-thinking, gemini-3.5-transcribe-live and gemini-3.5-live-translate-preview as current. The migration guide covers legacy models. Model names change, so check them against Google's docs.

Is it safe?

The skill folder bundles no scripts. SKILL.md has Claude run package installs (pip or npm) and fetch Google documentation. The plugin route also installs gemini-omni-flash-api, which has Python scripts. Keep your API key out of client-side code.

Information

updated Nov 2, 2026

Repository: google-gemini/gemini-skills
Folder: skills/gemini-live-api-dev

google-gemini/gemini-skills skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 3

gemini-live-api-dev SKILL.md · github.com
migration.md · github.com
gemini-skills README · github.com