madgraph

What problem does it solve?

This Skill provides a safe, framework-agnostic bridge to MadGraph 5_aMC@NLO, letting an AI agent drive non-interactive MG5 runs while capturing a compact cross-section summary and a reproducible run archive on disk.

Core Features & Use Cases

• Non-interactive MG driver: executes a single .mg5 script through a wrapper that logs stdout/stderr to a file and emits a concise structured summary.
• Cross-section extraction and diagram handling: reads the run banner or summary.txt to report xsec and uncertainties, and can generate aggregated Feynman-diagram PDFs from completed runs.
• Reproducibility and discovery: archives the exact script, stores a run_manifest.yaml with version, CLI args, and result, and supports per-run metadata for comparisons across runs.
• MG discovery and safety: detects MG installations, validates required Python toolchain, and guards user input with a slot-check before running.

Quick Start

Write a minimal MadGraph 5 script, run it with the wrappers, and inspect the produced run banner for the cross-section.