Skill UI

This skill provides implementation patterns for building robust and reliable AI voice agents and telephony automation systems using the Retell AI SDK. It covers best practices for handling common communication errors (like rate limiting or invalid API keys) and ensures stable interaction flow for voice-based applications and phone call automation scenarios.

This resource provides production-ready, structured patterns for building robust voice agent applications using the Retell AI SDK. It covers best practices such as implementing singleton client management, defining typed agent configurations with integrated LLMs, adding automatic retry logic for call failures, and managing large-scale outbound campaigns. Use this when building stable, scalable, and reliable voice automation systems.

This skill provides foundational security patterns and implementation guides for building AI voice agents and automating phone call workflows using Retell AI. It covers essential SDK usage (TypeScript), agent listing, and critical error handling (e.g., 401 Unauthorized, 429 Rate Limit) when integrating Retell into voice or telephony applications.

This skill provides comprehensive implementation patterns and detailed steps for migrating and upgrading Retell AI integrations. It is designed for developers building advanced AI voice agents, phone call automation, and complex telephony workflows using the Retell SDK. It covers setup procedures, best practices, and standard error handling for robust deployment.

A comprehensive, model-agnostic toolkit for advanced LLM development. It enables iterative prompt optimization through static analysis (token usage, clarity), measures critical RAG metrics (relevance, faithfulness, coverage), and validates complex agent workflows and configurations. Essential for rigorous LLM quality assurance and cost management.

SGLang is a high-performance serving framework for LLMs and VLMs that uses RadixAttention prefix caching to deliver structured outputs (JSON/regex/grammar) and agentic workflows with function calls, providing up to 5× faster inference than vLLM across multi-GPU production deployments.

Provides a structured workflow for defining agent-agnostic subagents with isolated context, covering purpose definition, metadata configuration, prompt writing, and common specialist patterns for verification, debugging, security auditing, and review.

This skill provides advanced guidelines for designing robust and effective tools specifically for autonomous AI agents. It shifts the focus from traditional developer APIs to contracts optimized for Language Models (LLMs). Key topics covered include treating tool descriptions as prompt engineering, implementing namespacing, applying the Consolidation Principle, and utilizing Architectural Reduction patterns (like file system access) to maximize agent reliability and workflow complexity.

torchforge is Meta's advanced PyTorch-native library designed for agentic Reinforcement Learning (RL). It fundamentally separates core RL algorithms from complex distributed infrastructure concerns. This allows researchers to rapidly prototype and experiment with novel RL methods (like GRPO) using clean, PyTorch-native abstractions, while automatically handling large-scale tasks like distributed training across thousands of GPUs using Monarch and TorchTitan.

Scaffolds production-ready generative AI agents on Vertex AI with Gemini models, multi-modal tools, RAG, and Cloud Run/Endpoint deployment, including observability and evaluation for enterprise scenarios.

This skill thoroughly inspects and validates Vertex AI Agent Engine deployments. It assesses critical components such as the Code Execution Sandbox, Memory Bank, A2A protocol compliance, security posture, and performance metrics. The tool generates weighted production-readiness scores (0-100%) and provides detailed, actionable recommendations to ensure the deployment is robust, secure, and ready for production use.

This skill guides the development of advanced voice agents, representing the frontier of natural human-AI interaction. It covers both low-latency Speech-to-Speech (S2S) architectures (like OpenAI Realtime API) and fully controllable STT→LLM→TTS pipelines. Key focus areas include minimizing end-to-end latency (targeting <800ms), handling conversational nuances, background noise, and managing interruptions for applications like customer support, IVR, and live translation.