This skill provides documentation for running prefill sensitivity analysis to measure how easily AI models can be manipulated into generating exploit code. Researchers use it to compare token-count versus logprob metrics for predicting reward hacking susceptibility across model checkpoints.

Finding qualified leads takes hours of research and guesswork. This skill analyzes your product and target market to identify companies that match your ideal customer profile, then provides prioritized leads with personalized contact strategies.

Understanding competitor advertising strategies helps you create more effective campaigns. This skill provides a structured approach to researching and analyzing competitor ads from public ad libraries to discover what messaging and creative approaches resonate with your target audience.

Finding accurate, up-to-date information online can be time-consuming. The Exa skill uses neural search technology to deliver high-quality, relevant web results through simple API calls.

Developers waste hours scrolling through GitHub repositories trying to find relevant issues or solutions. This skill provides structured search strategies, syntax examples, and workflow guides to quickly locate issues, pull requests, discussions, and code examples across any repository.

Understanding tool interfaces is critical for AI agent framework design. This skill analyzes how frameworks model, register, and execute tools with proper error handling. Use it to evaluate schema generation approaches, trace error feedback loops, and compare tool interfaces across frameworks.

Agent frameworks vary widely in error handling and recovery capabilities. This skill provides systematic methods to trace error propagation, evaluate sandboxing effectiveness, and assess production readiness before deployment.

Understanding how agents coordinate is critical for building scalable multi-agent systems. This skill provides frameworks and templates to analyze coordination models, handoff mechanisms, state sharing patterns, and communication protocols in any multi-agent architecture.

Agent frameworks struggle with context overflow and memory management. This skill provides systematic analysis of context assembly, eviction policies, and token budgeting strategies across different implementations.

Understanding how agent frameworks communicate with LLMs is essential for debugging, integration, and building new systems. This skill provides structured analysis methods to examine message formats, tool encoding, streaming mechanics, and multi-provider abstraction patterns.

Understanding how agent frameworks implement reasoning requires tracing complex control flows. This skill provides pattern signatures and extraction methods to identify reasoning topologies like ReAct, Plan-and-Solve, Reflection, and Tree-of-Thoughts from framework source code.

Understanding how existing AI agent frameworks work is crucial for building better systems. This skill provides a systematic methodology to analyze framework architectures, extract reusable patterns, and create reference implementations based on best practices from multiple frameworks.

Stop wasting time with random web searches. This skill provides a proven methodology to plan, execute, and synthesize web research efficiently. It breaks complex research into clear steps so you get comprehensive results faster.

Finding relevant research papers across physics, mathematics, computer science, and quantitative biology is time-consuming. This skill provides direct access to the arXiv preprint repository to search and retrieve papers with their titles, abstracts, and links.

Need to find real-time information or code examples for your coding tasks. This skill connects to Exa's search API to retrieve web content and programming context directly from your CLI.

Finding purchasable compounds for drug discovery is time-consuming and requires specialized database knowledge. This skill provides complete documentation for accessing the ZINC database with 230M+ compounds, enabling efficient virtual screening, similarity searches, and 3D structure downloads for molecular docking studies.

Finding patent and trademark information requires navigating multiple USPTO systems. This skill provides unified access to PatentSearch, PEDS, TSDR, and assignment databases for comprehensive IP research.

Predict molecular properties, generate new compounds, and analyze proteins using TorchDrug ML library. Train graph neural networks on molecules, proteins, and biomedical knowledge graphs for drug discovery and computational chemistry tasks.

Statistical analysis skills are essential for research but require proper methodology. This toolkit provides comprehensive guidance on hypothesis testing, regression, Bayesian methods, and APA reporting to help researchers conduct rigorous, reproducible analyses.

Complex systems with queues and shared resources are difficult to analyze without simulation. SimPy provides a Python-based framework for modeling processes, resources, and time-based events to predict system behavior before implementation.