What problem does it solve? Building AI features (chat, voice, vision, image generation, function calling) in Lens Studio for Specs requires correctly installing the Remote Service Gateway package, generating provider tokens, wiring scene hierarchies, and avoiding undocumented pitfalls like wrong Gemini model IDs and broken ASR event names. This Skill encodes all of that operational knowledge so the agent sets up working AI integrations without trial and error. ## Core Features & Use Cases - Zero-friction authentication: Programmatically generates all three provider tokens (SNAP, OPENAI, GOOGLE) via ExecuteEditorCode and wires them to the RemoteServiceGatewayCredentials component, with a UI fallback path. - Canonical assistant scripts: Provides copy-ready AIAssistant, GeminiAssistant, and OpenAIAssistant TypeScript implementations covering chat completions, Gemini Live, OpenAI Realtime, ASR hold-to-talk, and function calling. - Scene hierarchy automation: Supplies an idempotent ExecuteEditorCode bootstrap that builds the Requirements infrastructure (credentials, MicrophoneRecorder, DynamicAudioOutput) and wiring maps for all @input fields. - Use Case: A user asks for a voice-controlled AI assistant on Specs. The agent installs RSG 0.2.0+, generates tokens, copies the canonical scripts, builds the scene hierarchy, and verifies the ScriptComponents exist — producing a working hold-to-talk assistant with no manual wiring. ## Quick Start Set up a voice AI assistant in my Lens Studio project using the Remote Service Gateway with Gemini Live and hold-to-talk ASR.
