Skill UI

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 skill guides users through the entire lifecycle of developing AI agent skills. From initial concept (Understand) to structured design, writing the core skill logic, rigorous testing, and final performance optimization. It ensures skills are robust, accurate, and highly triggerable across any domain, including engineering, content creation, and research.

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.

Provides comprehensive guidelines for designing, auditing, and documenting empirical evidence for high-level AI research submissions (e.g., AAAI). It covers essential methodologies including strong baselines, single-factor ablations, robustness testing (shift, seed, prompt), statistical significance, and ensuring perfect alignment with reproducibility checklists and ethical reporting. Use this guide before submission to ensure empirical evidence strongly supports all claimed AI contributions.

Provides comprehensive guidelines for designing, auditing, and reporting high-rigor multi-agent experiments, especially in game theory and self-play systems. It emphasizes testing the *interaction* claim—such as convergence to equilibrium, truthful behavior, or emergent cooperation—against held-out opponents, populations, and strategic deviations, moving beyond simple single-agent leaderboard scores.

A comprehensive guide for researchers to determine if their work represents a genuine cross-modal or media-systems contribution suitable for ACM Multimedia (ACM MM). It helps route projects by testing for cross-modality, media system advances, or human-centric results, distinguishing between pure CV, NLP, or system-heavy claims.

A comprehensive guide for structuring the empirical evidence section of top-tier NLP conference papers (e.g., EACL). It mandates rigorous controls, including fair baselines (tuned models, explicit LLM prompts), statistical significance testing (variance, CI), contamination checks, and detailed human evaluation protocols. Use this to elevate your research from mere assertion to scientifically audited proof.