Skill UI

Extends transformer context windows using RoPE, YaRN, ALiBi, and position interpolation so long documents (32k-128k+ tokens) can be processed, pre-trained models lengthened, and efficient positional encodings or extrapolation strategies deployed.

This skill provides methodologies for designing sophisticated, layered memory architectures for AI agents. It moves beyond simple context windows and basic vector stores by incorporating short-term, long-term, and graph-based memory systems. Implementing this is crucial for agents that require persistence across sessions, maintaining complex entity consistency, and performing advanced, relationship-driven reasoning over accumulated knowledge bases, including time-aware fact tracking.

This skill provides advanced strategies for managing large language model (LLM) context windows and token usage via OpenRouter. Functions include context window lookup, context-aware model selection, conversation trimming, document chunking for RAG, and prompt caching. Use this when dealing with long conversations or large documents to maximize quality while minimizing API costs and preventing context limit overflow.

Detect malicious scheduled task creation and modification using Sysmon (Event IDs 1/11), Windows Security events (4698/4702), and correlation of suspicious parents, public paths, and encoded arguments for persistence and lateral movement analysis.

Monitor and analyze Windows Management Instrumentation event subscriptions (filters, consumers, bindings) to detect and investigate fileless persistence mechanisms and attacker activity triggered via WMI, especially when other persistence traces are absent.

Guide for building passwordless authentication via Microsoft Entra ID, covering FIDO2 keys, Windows Hello for Business, Microsoft Authenticator passkeys, and conditional access policies for phishing-resistant MFA deployments.

This skill provides real-time object detection capabilities by leveraging the Google Coral Edge TPU accelerator. It operates natively within Windows Subsystem for Linux (WSL), allowing high-speed, low-latency inference on live camera frames. By utilizing hardware acceleration, it achieves impressive performance (e.g., ~4ms inference at 320x320), making it ideal for edge computing projects requiring rapid visual analysis and resource efficiency.