Scientific Computing Pipeline Reproducibility Audit Plan

Role: Principal Research Software Engineer specializing in computational reproducibility and open science.

Context

• Institution: {{research_institution}}
• Domain: {{scientific_discipline}}
• Codebase: {{target_repository}}
• Execution Environment: {{compute_environment}}
• Data Standard: {{data_provenance_standard}}
• Target Milestone: {{publication_deadline}}

Task

Design a structured reproducibility code review plan that guides peer reviewers and lab members through auditing scientific algorithms, fixed random seeds, package dependencies, and raw data transformations in {{target_repository}} prior to {{publication_deadline}}.

Method

1. Establish baseline environment isolation requirements tailored to {{compute_environment}}.
2. Outline source code inspection checkpoints for deterministic random seed management and floating-point stability.
3. Formulate verification criteria for data input pipelines against {{data_provenance_standard}}.
4. Define script modularity and unit test coverage requirements for core numerical routines.
5. Design an artifact verification protocol covering configuration files, lockfiles, and container recipes.
6. Schedule staged review milestones mapping directly to {{publication_deadline}}.
7. Detail failure-triage procedures when benchmark outputs deviate from expected figures.

Constraints

• Focus strictly on technical reproducibility, computational correctness, and code hygiene.
• MUST include explicit validation checkpoints for deterministic random states and hardcoded file paths.
• MUST NOT require proprietary third-party commercial review tooling unavailable in open academic environments.
• Limit recommendations to pragmatic steps achievable by academic researchers.
• Reference {{scientific_discipline}} domain norms for code documentation.

Output format

Provide the review plan using these exact sections:

1. Executive Audit Charter (1 paragraph summarizing objectives and scope)
2. Environment & Dependency Review Matrix (bulleted audit criteria)
3. Numerical Correctness & Provenance Protocol (numbered review steps)
4. Staged Timeline & Deliverable Schedule (phased breakdown leading to {{publication_deadline}})

Self-review

• Confirm all 6 variables are referenced naturally in the text.
• Verify that the plan explicitly addresses computational reproducibility in {{compute_environment}}.
• Ensure the format adheres to the four designated output sections without omitted parts.