Skill UI

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 automates the scholarly process of refining a generated academic paper. It executes an iterative loop (Review → Fix → Recompile) using advanced LLMs (like GPT-5.5) over multiple rounds. It is designed not for primary research, but specifically for polishing writing quality, correcting theoretical inconsistencies, improving presentation, and enhancing narrative flow after the initial draft is complete. It helps elevate a paper from a draft to a polished submission.

This skill facilitates autonomous, multi-round research review and refinement. It continuously submits work for assessment via external reviewers (Codex or manual), implements suggested fixes, and re-reviews until the designated positive threshold is met or the maximum number of rounds is reached. It manages state persistence and adjusts difficulty levels for rigorous, iterative quality assurance.

This skill automates the research improvement process by creating a closed loop: Review -> Implement Fixes -> Re-review. It leverages any OpenAI-compatible LLM API to iteratively refine research work until a positive assessment is reached or a maximum round limit is hit. Ideal for academic or technical project refinement.

This advanced skill automates the rigorous, iterative review and refinement of complex research topics or papers. It runs multi-round cycles, using an external LLM (MiniMax API) to provide continuous, constructive feedback. The process guides the user through review, fixes, and re-review stages until a positive assessment is achieved or the maximum round limit is reached. It includes robust state persistence for long-running academic workflows.

A powerful academic tool designed for researchers. Given a broad research direction, it systematically surveys the current literature (landscape survey), identifies knowledge gaps, and generates multiple concrete, high-potential hypotheses. It then validates and ranks these ideas to propose novel, publishable research projects.

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.

Produces structured proof packages for ML/AI theory claims by extracting statements, assumptions, and sketches, then drafting a dependency-aware proof or blockage report ensuring every step is justified before updating PROOF_PACKAGE.md.

This skill automates the end-to-end workflow for deploying and executing ML training and inference experiments. It intelligently detects the target environment (Local GPU, Remote SSH, Vast.ai, Modal serverless) and handles necessary steps like GPU pre-checks, code synchronization (rsync/git), integrating ML tracking tools (W&B), and launching the compute job via screen sessions or serverless deployment. Use when initiating a complex training job.

This module provides a centralized interface for the AgentHub message board. It allows multiple AI agents and the coordinator to communicate, track tasks, and log results through structured markdown posts across dedicated channels (Dispatch, Progress, Results). It is essential for monitoring complex multi-agent workflows and ensuring transparent collaboration and auditing.

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 tool audits the token consumption across all components (agents, skills, rules, MCP servers) within a Claude Code session. It analyzes token overhead, identifies bloat (e.g., overly long descriptions, redundant components), and generates a detailed context budget report. Use it to maintain optimal context window usage, improve performance, and prevent context saturation.