nash-run

What problem does it solve?

This Skill helps you execute reproducible game-theory simulations and experiments so you can obtain machine-readable results instead of running ad-hoc, non-repeatable trials.

Core Features & Use Cases

• Reproducible simulation runs: Executes NASH CLI commands to run preset game environments with controlled seeds and saved JSON outputs.
• Parameter sweeps & multi-model comparisons: Splits sweeps across configs (including parallel subagent splitting) and runs multiple environments in parallel for side-by-side evaluation.
• Built-in post-run validation and visualization: Triggers validate and viz steps after runs so results are checked and report-ready.

Real examples

You can run a Hawk-Dove simulation across multiple seeds, sweep a key parameter range, or compare several environments (e.g., public_goods vs common_pool) and then produce validation outputs and charts.

Quick Start

Ask the AI agent to run a 200-round Hawk-Dove simulation with 100 agents, seed 42, and save the results to results.json, then validate and visualize the output.