# Paperlab Reproducibility Capsule

> Build PaperLab reproducibility capsules with raw data, scripts, notebooks, command logs, environment records, seeds, manifests, and rerun instructions.

- Skill: `equinor/paperlab-reproducibility-capsule` (Agent Skill)
- Install (CLI): `npx skillmds@latest add equinor/paperlab-reproducibility-capsule`
- Raw SKILL.md: https://api.skillmd.com/api/skills/equinor/paperlab-reproducibility-capsule/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: equinor (https://skillmd.com/u/equinor)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/equinor/paperlab-reproducibility-capsule

---


# PaperLab Reproducibility Capsule

## When to Use

USE WHEN: preparing a paper submission, revision, supplementary material, or
book computational appendix that must be rerunnable.

## Capsule Structure

```text
replication_package/
  README.md
  manifest.json
  commands.ps1
  environment.txt
  raw_data/
  scripts/
  notebooks/
  expected_outputs/
```

## Manifest Fields

- NeqSim commit hash,
- PaperLab commit hash,
- Java version,
- Python version,
- dependency list,
- random seeds,
- input data files,
- output files and tolerances,
- commands required to rerun,
- known platform limitations.

## Pass Criteria

- A reviewer can rerun the core results without reading private notes.
- Figures and summary metrics have regeneration commands.
- Non-public data dependencies are clearly excluded or replaced with fixtures.
- Expected numerical tolerances are stated.

## Safety Rules

- Do not include internal documents, credentials, tokens, or browser profiles.
- Do not rely on local absolute paths in rerun commands.
- Keep large generated files out of the capsule unless required for review.