Skill UI

Stable Baselines3 is a robust, PyTorch-based library providing reliable implementations of state-of-the-art reinforcement learning algorithms (PPO, SAC, DQN, etc.). It is designed for standard RL experiments, quick prototyping, and developing complex single-agent systems. The library supports training agents, creating custom Gymnasium environments, and running vectorized simulations for accelerated learning.

A robust job queue designed to orchestrate large-scale, multi-seed, and multi-configuration Machine Learning experiments on remote GPU clusters. It handles complex workflows, including wave transitions, dependency chaining (teacher-student), Out-of-Memory (OOM) retries, and stale screen cleanup, ensuring reliable execution for demanding scientific computing tasks.

This structured mode acts as an experiment design partner for Product Managers. It guides users through formulating testable, falsifiable hypotheses, designing rigorous A/B tests (including sample size, duration, and guardrails), and interpreting results honestly. It helps PMs move from vague beliefs to clear, data-backed decisions: Ship, Iterate, or Kill.

This tool functions as a post-task epilogue, designed to audit and record the entire research journey. It scans the conversation history at the end of a coding or research session to extract critical research artifacts, including decisions, experiments, dead ends, claims, heuristics, and major pivots. It systematically writes this provenance data into the structured `ara/` directory, ensuring a faithful and auditable trace of how the project truly evolved.

This skill automates the complex process of resubmitting an already polished academic paper to a different conference or journal. It is designed for scenarios where the core content, experiments, and bibliography must remain untouched, but the paper needs micro-edits to accommodate new venue constraints, page limits, or prior reviewer feedback. It ensures the new submission is generated as a separate, sibling directory to maintain the integrity of prior submissions.

This skill guides the safe execution of chaos engineering experiments to surface weaknesses in production systems without causing actual outages. It provides tools to design robust experiments by defining hypotheses, calculating the potential 'blast radius' before failure injection, setting strict abort criteria, and generating structured postmortems. Ideal for SRE teams running game days and resilience tests.

This guide provides a comprehensive framework for designing statistically rigorous A/B tests, split tests, and multivariate experiments. It covers everything from formulating strong hypotheses and calculating sample sizes to selecting appropriate primary, secondary, and guardrail metrics for various business contexts.

This expert framework guides users through the systematic process of conducting rigorous A/B tests and growth experiments. It covers all stages: formulating strong hypotheses, selecting appropriate metrics (primary, secondary, guardrail), calculating necessary sample sizes, and choosing the best implementation method (client-side vs. server-side). Use this when planning to compare two versions or build a systematic growth experimentation program.

This tool guides users through the critical process of designing experiments *before* data collection. It helps select appropriate structures—such as factorial, crossover, or blocked designs—to ensure that collected data is interpretable and allows for robust causal inference. Use it when planning scientific studies, assigning subjects, or optimizing responses across multiple factors.