Minimal Run And Audit

You call this after you've already figured out what command to run in a deep learning reproduction attempt and now need clean evidence capture. It executes your smoke test or documented inference run, then writes standardized reports to repro_outputs including patch notes if files changed and a scientific changelog when changes touch metrics or evaluation logic. The boundaries are strict: it won't pick your reproduction target, won't do paper analysis, and won't hide risky edits. Think of it as the auditable execution layer that sits between "I know what to try" and "here's normalized proof of what happened." Good for turning messy README-driven repos into something you can actually cite evidence from.

skills.sh installs: 492.4K
GitHub stars: 486
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add lllllllama/rigorpilot-skills --skill minimal-run-and-audit --agent claude-code

Installs into .claude/skills of the current project.

Files

minimal-run-and-audit

Use this as the Rigor Run skill. The installed slug remains minimal-run-and-audit for compatibility.

Use the shared operating principles in ../ai-research-reproduction/references/agent-operating-principles.md; this skill should make run evidence auditable without turning every command into a rigid protocol.

When to apply

After a reproduction target and setup plan exist.
When the main skill needs execution evidence and normalized outputs.
When a smoke test, documented inference run, documented evaluation run, or other short non-training verification is appropriate.
When the user already knows what command should be attempted and wants execution plus reporting only.

When not to apply

During initial repo scanning.
When environment or assets are still undefined enough to make execution meaningless.
When the task is a literature lookup rather than repository execution.
When the user is still deciding which reproduction target should count as the main run.