Skill UI

This skill provides comprehensive guidance for building production-grade MLOps pipelines. It covers the entire machine learning lifecycle, from initial data ingestion and rigorous feature engineering to model training, advanced validation (A/B testing, performance drift), and robust deployment strategies (Canary, Blue-green). It enables the design and orchestration of scalable, reproducible, and observable ML systems using tools like Airflow and Kubeflow.

Hypogenic automates LLM-driven hypothesis generation and testing over tabular datasets, blending literature insights with empirical analysis to accelerate discovery in domains like deception detection or AI content evaluation.

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.

Phoenix Evals enables developers to build and run evaluators for AI and LLM applications. It supports both code-based and LLM-based evaluation methods, including RAG systems. Users can create datasets, run experiments, validate evaluator accuracy, and implement production guardrails. Available in Python and TypeScript, it helps ensure model performance and reliability through systematic testing and human validation.

A comprehensive security auditing framework designed to systematically detect sensitive information disclosure from Large Language Models (LLMs). It uses multi-phase, escalating dialogue probes to test for leaks such as system prompts, API keys, PII, and internal configurations. Essential for red teaming and rigorous model security assessment.

This minimal working example demonstrates Abridge's ambient AI clinical documentation capabilities. It simulates a complete patient encounter session, allowing users to submit transcripts and receive a highly structured clinical note. Ideal for testing EHR connectivity, verifying AI workflow, or learning how structured medical data (SOAP notes, ICD/CPT codes) is automatically generated from free-text conversation.

This harness provides comprehensive, reproducible evaluation pipelines for complex LLM chains and agents (LangChain/LangGraph 1.0). It integrates golden dataset management, LangSmith evaluation, RAGAS metrics, deepeval LLM-as-judge, and structured agent trajectory analysis. Use it when establishing quality benchmarks for new chains, diagnosing performance regressions after model switches, or implementing CI/CD gates to prevent quality drops.

This comprehensive benchmark suite evaluates the performance of Vision-Language Models (VLM) on complex video anomaly detection tasks using the SmartHome-Bench dataset. It covers a wide range of real-world smart home scenarios, including security breaches, elderly fall detection, and unusual pet behavior. Unlike single-frame analysis, this benchmark requires deep multi-frame video understanding, making it a robust tool for testing the advanced AI capabilities of video monitoring systems.

A comprehensive expert playbook detailing advanced Large Language Model (LLM) prompt injection techniques. Covers direct methods (instruction override, role play), indirect attacks (RAG poisoning, web data manipulation), and sophisticated tool/function calling abuse for data exfiltration and security bypass. Essential for rigorous AI application security testing.

This skill provides a repeatable assessment for RAG systems, testing for critical vulnerabilities in vector stores and embedding models. It specifically addresses embedding inversion, cross-tenant data leakage, and knowledge-base poisoning, aligning with OWASP LLM08:2025. Use when performing authorized security audits of retrieval layers and multi-tenant isolation.

This skill provides comprehensive techniques for security-testing Retrieval-Augmented Generation (RAG) pipelines. It probes for indirect prompt injection, which occurs when malicious instructions are embedded in retrieved documents, and tests for embedding-space manipulation. Using advanced tools like NVIDIA garak, Promptfoo, and PyRIT against vector stores (FAISS, Chroma, etc.), users can validate retrieval guardrails and secure CI/CD gates against adversarial inputs and data leakage.

This tool enables complex web interactions using plain natural language commands. Instead of relying on unreliable selectors, it utilizes a powerful page-agent and LLM loop to interpret user intent, performing actions like filling forms, clicking buttons, and extracting data from dynamic websites. Ideal for testing and automation in pages with ambiguous or changing markup.