Skill UI

Best practices for building AI chat interfaces with Makepad using MolyKit, covering cross-platform async, OpenAI/LLM integration, SSE streaming, chat widgets, and bot clients.

Unified LLM inference layer built into Neon branches: one credential and one OpenAI-compatible API for frontier and open-source models from seven providers. Switch models without rewiring, wire it into OpenAI/Anthropic SDKs, and keep AI requests branch-scoped.

This blueprint guides the creation of F.R.I.D.A.Y., a tactical, low-latency voice assistant inspired by Iron Man. It utilizes the specialized Pipecat framework to orchestrate a comprehensive, real-time audio pipeline: STT (OpenAI Whisper/GPT-4o) -> LLM (Gemini) -> TTS (OpenAI Nova). Ideal for developers needing to build advanced, multi-provider, conversational voice applications.

PydanticAI is a robust Python framework that integrates Pydantic's type-checking and validation capabilities into Large Language Models (LLMs). It allows developers to build production-ready AI agents that guarantee structured, validated outputs, moving beyond simple raw strings. Key features include supported tool calling, dependency injection for testability, and multi-model compatibility across major providers like OpenAI, Anthropic, and Gemini. Use this when reliability and correct data structure are paramount.

RouterBase acts as a unified, OpenAI-compatible API gateway, allowing applications to route requests across diverse large language models (LLMs) like GPT, Claude, and Gemini. It helps developers migrate existing OpenAI SDK calls, implement sophisticated primary/fallback model selection based on cost, latency, and quality, and manage multimodal payloads (image, audio, video) from a single, standardized endpoint.

Simulates Sam Altman, CEO of OpenAI and former Y Combinator president, to provide insights on startups, AGI, fundraising, and the future of AI. Use for advice in the style of Sam Altman.

Deploy and manage vLLM for high-throughput LLM inference. Configure continuous batching, tensor parallelism, quantization, and OpenAI-compatible API endpoints for production LLM serving.

Voice agents enable natural human-AI conversation through speech recognition and synthesis. This skill covers low-latency architectures like speech-to-speech pipelines and STT-LLM-TTS workflows. It includes optimization strategies for sub-800ms response times, voice activity detection, and integration with tools like Deepgram, ElevenLabs, and OpenAI. Ideal for building production-ready voice assistants with natural interaction flows.

Expert in building voice AI applications - from real-time voice agents to voice-enabled apps. Covers OpenAI Realtime API, Vapi, Deepgram, ElevenLabs, LiveKit, and WebRTC. Focuses on low-latency, production-ready voice experiences.

This module adopts the persona, vocabulary, and critical pedagogical style of Yann LeCun. It provides deep insights into open-source technology, the concept of technological sovereignty, and the geopolitical incentives driving major AI companies (Meta, OpenAI, Google). It structures explanations using Socratic methods, grounding complex AI concepts in physical analogies.