google-gemini/gemini-skills
3 skills33.8k installs11.6k starsGitHub
Install
npx skills add https://github.com/google-gemini/gemini-skillsSkills in this repo
1Gemini Api DevGemini API Development teaches developers how to build applications using Google's Gemini models across multiple programming languages. It covers SDK setup for Python, JavaScript/TypeScript, Go, and Java, along with model selection, multimodal content handling, function calling, and structured outputs. Use this skill when integrating Gemini models into backend services, APIs, or agent systems. Includes current model specifications and best practices for production use.17.5kinstalls2Gemini Interactions ApiGemini Interactions API is the current, recommended interface for using Gemini models and managed agents. Developers use it to generate text with gemini-3.5-flash or gemini-3.1-pro, conduct multi-turn conversations with state stored server-side, invoke managed agents like Antigravity (code execution, web browsing) or Deep Research, stream incremental responses, call functions and tools (Google Search, code execution, URL context, file search), and output structured JSON or images. The SDK replaces the deprecated generateContent API. Core workflows include creating interactions with model/agent/input, chaining turns via previous_interaction_id, streaming with step.delta events, polling background tasks, and building custom agents with sandboxed Linux environments.9.8kinstalls3Gemini Live Api DevThe gemini-live-api-dev skill teaches low-latency, real-time voice and video applications with the Gemini Live API over WebSockets. It documents bidirectional audio streaming, camera or screen video frames, in-session text, input and output transcriptions, voice activity detection, native audio with configurable thinkingLevel, synchronous function calling, Google Search grounding, session compression and resumption, ephemeral tokens for browser auth, and live translation across seventy plus languages. Coverage targets google-genai Python and @google/genai JavaScript SDKs, recommends gemini-3.1-flash-live-preview, explains PCM audio formats, sendRealtimeInput versus sendClientContent rules, interruption queue handling, partner WebRTC integrations via LiveKit or Pipecat, and migration off deprecated 2.5 models. The skill also details connection lifetime limits, modality constraints, audioStreamEnd flushing, processing all server event parts, and search_docs MCP lookup when available. Invoke when wiring mic-to-speaker agents, multilingual interpreters, or vision-aware live sessions. The reference also notes partner WebRTC stacks, GoAway session lifetime handling, and deprecated model.6.5kinstalls