Skill UI

This tool is designed to help optimize and refine prompts used with Large Language Models (LLMs). By structuring inputs, it ensures that the instructions given to AI systems are clear, comprehensive, and contextually rich, resulting in significantly higher quality and relevance of the generated output. Useful for developers, writers, and researchers.

This tool provides a comprehensive framework for analyzing web server logs (Apache/Nginx) to detect various security threats. It uses advanced regex matching against OWASP attack signatures to identify SQL injection, Local File Inclusion, and Directory Traversal attempts, alongside behavioral analysis for brute-force attacks and web scanner fingerprints. The system also enriches data using GeoIP for source attribution, generating a prioritized report for SOC analysts.

This skill implements an autonomous, multi-round research review loop, simulating the rigorous peer-review process of top-tier conferences. It iteratively reviews, identifies weaknesses, suggests fixes, and re-reviews a research document using the MiniMax API. The system manages state persistence across multiple rounds, ensuring continuous improvement until a 'ready' assessment is achieved or the maximum round limit is reached. Ideal for academic writing, thesis refinement, or comprehensive project documentation.

This skill generates and ranks publishable research ideas based on a broad direction provided by the user. It systematically validates concepts through pilot experiments and model reviews, optimizing for GPU efficiency and novelty. Suitable for brainstorming sessions, academic exploration, and identifying workable research paths.

A comprehensive guide and workflow demonstrating how to use YARA rules for proactive threat hunting. This method helps detect unknown malware, suspicious patterns, and Indicators of Compromise (IOCs) across filesystems, memory dumps, and network artifacts, making it essential for incident response and security analysis.

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.

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.

Designs systematic ablation studies for ML paper submissions. Helps isolate component contributions, test hyperparameters, and answer reviewer questions. Codex plans experiments while CC implements them, ensuring rigorous validation before submission.

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 framework implements robust guardrails for LLM applications, validating both user inputs and model outputs. It detects prompt injection, sensitive PII (Personally Identifiable Information), toxic content, and ensures structured output compliance using advanced tools like NVIDIA NeMo Guardrails and Guardrails AI. Essential for building secure, reliable chatbots and RAG systems.

Assesses AI/ML systems for prompt injection, jailbreak, model inversion, data poisoning, and agent tool abuse using MITRE ATLAS mapping, signature detection, and scoring metrics before deployment.

This harness transforms standard code environments into fully persistent, autonomous agent systems. It leverages native features like scheduled tasks (Crons), remote dispatching, and structured memory storage (MCP) to enable continuous, self-directing automation. Ideal for building AI assistants that require long-term context, recurring workflows, and complex, multi-step operations, eliminating the need for external agent frameworks like AutoGPT.