Skill UI

This guide details comprehensive physical security penetration testing techniques, essential for red teaming engagements. It covers methods such as tailgating, badge cloning (RFID), lock bypassing, and deploying rogue network devices (e.g., Pineapple, LAN Turtle). It provides practical toolkits and MITRE ATT&CK mapping to evaluate an organization's physical and digital control weaknesses.

A specialized tool for formalizing and verifying complex mathematical proofs, particularly in Machine Learning and AI theory. It guides users through structuring theorems, identifying explicit assumptions, performing feasibility triage, and ensuring every logical step is rigorously justified, preventing the fabrication of proofs.

A powerful tool for conducting comprehensive academic research and literature reviews. It searches, analyzes, and synthesizes key ideas from multiple sources, including local PDF libraries (Zotero, Obsidian), the general web, and specialized academic databases (Semantic Scholar, DeepXiv). Use this when you need to map the historical context, identify related works, or understand complex concepts across multiple studies on a given topic.

Get a deep, multi-round, and adversarial critical review of your research work, ideas, or drafts. This skill simulates the perspective of a senior ML reviewer (e.g., NeurIPS/ICML level) using dedicated external backends (Codex or Manual). It is designed to find logical gaps, missing experiments, and narrative weaknesses, providing brutal honesty to elevate your academic contribution.

This tool allows users to systematically evaluate and rank the outputs of multiple agents within an AgentHub session. It supports metric-based evaluation (running custom commands for measurable metrics like speed or resource usage), or an advanced LLM Judge Mode. The LLM Judge compares results based on criteria like correctness, simplicity, and overall quality, helping users determine the best-performing agent for the task.

Soft Predict Future is a structured foresight engine that triggers whenever the user asks about future outcomes or scenarios. It enforces a nine-step analytical pipeline (validation, signal collection, scoring, driver extraction, etc.), highlights required searches and inputs, and directs the user to web-enabled reasoning for predictive analysis.

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.

This skill systematically probes an AI model's core behavioral patterns across six critical dimensions, including reasoning, refusal, formatting, and grounding. It generates a comprehensive, styled HTML report with radar charts and actionable insights, allowing users to understand a model's true capabilities, compare different models, or perform compliance and red-team assessments without needing an external API key.

AKF is the native metadata format specifically designed for tracking AI-generated or AI-modified content. It allows users to stamp files (including DOCX, PDF, images, and code) with detailed trust scores, source provenance, and compliance metadata. This tool is essential for professional workflows requiring strict auditing, compliance assurance, and verifiable data lineage, especially concerning regulations like the EU AI Act, HIPAA, and SOX.

This tool systematically designs and plans ablation studies crucial for advanced research paper submissions. It simulates the role of a rigorous ML reviewer, guiding the process of isolating the specific contributions of each novel component, testing hyperparameter sensitivities, and addressing potential reviewer questions. The workflow ensures that every experiment has a clear hypothesis, maximizing the scientific impact of the research results.

This skill detects AI-generated deepfake audio used in voice phishing (vishing) attacks. It functions by extracting critical spectral features (including MFCCs, spectral centroid, and zero-crossing rate) and passing them through advanced machine learning models (like Random Forest and Gradient Boosting). The system supports batch forensic analysis, provides confidence scores, and generates detailed reports to verify the authenticity of recorded voice calls, making it essential for cybersecurity incident response and fraud investigation.

This skill is designed for running comprehensive evaluations of Hugging Face Hub models on local hardware. It supports multiple evaluation frameworks, including `inspect-ai` and `lighteval`, and provides flexibility in backend selection (e.g., `vllm`, Transformers, or `accelerate`) to optimize for performance and compatibility. Use this for local GPU benchmarking, smoke testing, and backend selection, but note that it does not handle remote job orchestration or result publishing.