Computational Notebook Reproducibility Review Spec
Create a formal code review specification to audit scientific notebooks for computational reproducibility and data integrity.
Use this template when preparing code review standards for research software, Jupyter/R notebooks, or data science scripts prior to peer-reviewed publication.
Role: Senior Research Software Engineer and Scientific Reproducibility Auditor
Context
- Institution or Lab: {{institution_name}}
- Research Domain: {{research_domain}}
- Primary Programming Language: {{primary_language}}
- Execution Environment or Framework: {{notebook_framework}}
- Dataset Licensing & Access: {{dataset_licensing}}
- Target Journal / Venue: {{target_journal}}
Task
Draft a comprehensive code review specification document that establishes reproducibility, deterministic execution, and data provenance standards for scientific software submitted to {{target_journal}}.
Method
- Analyze the requirements of {{research_domain}} to establish baseline standards for computational determinism, random seed configuration, and numerical stability.
- Define strict dependency pinning and environment containerization review checkpoints suited for {{notebook_framework}} and {{primary_language}}.
- Formulate review criteria for raw dataset ingestion, data versioning, and compliance with {{dataset_licensing}}.
- Design validation rules for intermediate caching, pipeline modularization, and memory management during large-scale simulation runs.
- Establish automated unit testing, smoke testing, and artifact generation requirements within the review workflow for {{institution_name}}.
- Detail code documentation norms, including math formula attribution, parameter boundary comments, and figure generation linkage.
- Map triage rules for handling non-deterministic libraries, GPU acceleration variance, and legacy runtime deprecation.
Constraints
- All review criteria MUST explicitly prioritize end-to-end computational reproducibility over raw execution speed.
- Review criteria MUST NOT accept unpinned runtime dependencies or hardcoded local file paths.
- Include concrete pass/fail metrics for each inspection dimension.
- Tailor guidelines specifically to {{primary_language}} workflows within {{research_domain}}.
Output format
Provide the review specification structured under these exact section headers:
- Scope & Reproducibility Objectives (max 150 words)
- Environment & Dependency Audit Checklist (5-7 itemized rules)
- Data Pipeline & Determinism Verification Spec (5-7 itemized rules)
- Code Quality & In-Notebook Documentation Standards (4-6 inspection points)
- Review Pass/Fail Decision Matrix (table with criteria, severity, remediation)
Self-review
- Confirm all 6 variables ({{institution_name}}, {{research_domain}}, {{primary_language}}, {{notebook_framework}}, {{dataset_licensing}}, {{target_journal}}) are utilized logically.
- Ensure environment pinning and random seed enforcement are clearly articulated.
- Check that the total text adheres to the section structure without conversational filler.
Explicit role, a named task, and discrete steps the model can follow.
Background, inputs and variables the model needs before it starts.
Hard boundaries — what the model must and must not do.
A named, field-level shape for the response.
Ordered work items that force analysis before an answer.
Length and structure that travel across frontier models.
Signal density — instruction weight without padding.
Documented variables so the scaffold adapts to new inputs.
Quality bar, assumptions and behaviour when inputs are thin.
How much real usage the template has behind it.