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.

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.

This skill outlines the CoRL 2026 review pipeline, including the OpenReview double-anonymous system, reviewer/AC/SAC hierarchy, scoring thresholds (weak accept), and rebuttal mechanics. It helps authors understand leverage points like the first-round rejection gate and discussion windows to improve acceptance chances.

Use for live Windows or macOS GUI tasks that structured tools cannot complete. Discover approved apps and windows, observe current state, act via semantic elements or shortcuts, and verify every requested outcome. Includes an operating loop for reliable desktop automation and visual verification.

Looks up precomputed AlphaGenome Atlas effects for GRCh38 SNVs (AVI score, Phred, 18 SHAP attributions, track scores) and scores variants or scans windows on demand with the AlphaGenome model for human and mouse, plus builds Atlas deep links.

Elite AI context engineering specialist mastering dynamic context management, vector databases, knowledge graphs, and intelligent memory systems. Provides best practices for orchestrating complex AI workflows, optimizing context windows, and integrating enterprise knowledge bases.

Strategies for managing LLM context windows: tiered routing, summarization, trimming, token counting, and prompt placement to avoid context rot in multi-turn AI applications.

Provides read-only platform APIs for inspecting model management, pricing, and usage tracking. Use to verify account identity, region, endpoints, time windows, costs, and gaps without modifying keys or billing.

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.

Expert in building Retrieval-Augmented Generation systems for LLM applications. Masters embedding models, vector databases, chunking strategies, and retrieval optimization. Capable of designing retrieval pipelines, implementing hybrid search, and optimizing context windows to ensure high-quality generation from retrieved documents.