Skill UI

PufferLib is a high-performance library designed for advanced reinforcement learning simulations and training. It excels in parallel environment simulation, offering optimized vectorization and native support for multi-agent systems. Utilizing algorithms like PuffeRL (optimized PPO+LSTM), it achieves millions of steps per second, significantly accelerating research and deployment across single-agent and multi-agent environments.

Real-time research lookup that auto-selects between Parallel Chat API and Perplexity sonar-pro-search to gather papers, facts, market data, and cited findings.

Stable Baselines3 is a robust, PyTorch-based library providing production-ready implementations of core reinforcement learning algorithms (PPO, SAC, DQN, etc.). It facilitates the training of single-agent RL models, supports the creation of custom Gymnasium environments, and handles advanced parallel training via vectorized environments. It is ideal for rapid prototyping and rigorous RL experiments.

TileDB-VCF lets you ingest, store, query, and export large genomic VCF/BCF datasets using TileDB’s sparse arrays, supporting incremental sample growth, parallel regions queries, and cloud-local interoperability for population genomics workflows.

Zarr is a Python toolkit for chunked N-D array storage with compression, sharding, and native integrations with NumPy/Dask/Xarray, ideal for parallel I/O and scalable cloud pipelines.

A powerful tool for architecting production-grade multi-agent workflows. It allows users to define complex, multi-step AI pipelines using structured patterns (sequential, parallel, hierarchical). Key features include defining explicit handoff contracts, implementing robust failure handling, and managing context/cost discipline across long-running processes, ensuring reliable and scalable AI system deployment.

This powerful skill manages Git worktrees to enable safe, isolated parallel feature development. It standardizes branch isolation, automatically allocates unique network ports, and synchronizes environments, allowing multiple agents or terminal sessions to work on different branches simultaneously without conflicts. It also provides mechanisms for detecting and safely cleaning up stale worktrees.

Dispatching parallel agents lets you split independent investigations into focused prompts, coordinate isolated fixes concurrently, and review summaries for integration once each agent finishes.

Parallel Web Research skill orchestrates Parallel Chat and Extract APIs to search the internet, perform deep intelligence work, and craft cited summaries, with model controls, JSON outputs, and citation verification for market, news, technical, and general queries.

A comprehensive guide and methodology for identifying, measuring, and fixing performance bottlenecks across the entire stack. This skill covers optimizing database queries (e.g., N+1 problems, missing indexes), improving API efficiency (caching, parallel processing), and enhancing frontend performance (memoization, code splitting, image optimization). It emphasizes the critical 'Measure First' principle to ensure all optimizations are data-driven and effective.

Coordinates a pre-task intelligence protocol that activates all relevant agents, synthesizes parallel analyses, estimates real execution time, maps probable risks, and produces a contingency-aware plan before executing the user’s request.

Provision enterprise-grade Playwright Java tests with a Page Object Model scaffold, JUnit 5 workflows, thread-safe parallel runs, Allure reporting, CI-friendly setup, and hybrid API/UI coverage to fix flakes and scale browsers.