repro-record-workflow

Automate per-subtask logging into CODEX_RECORD.md for reproducible research workflows.

Updated Apr 10, 2026
One-click install
npx skills add https://github.com/KwongFuk/codex-skills --skill repro-record-workflow
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: repro-record-workflow
Source: https://github.com/KwongFuk/codex-skills/tree/main/global/repro-record-workflow
Command: npx skills add https://github.com/KwongFuk/codex-skills --skill repro-record-workflow

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

Reproducibility in research projects often requires manual logging of subtask steps and outcomes across multiple tools. This Skill couples reproduction work with a dedicated CODEX_RECORD.md to ensure traceability.

Core Features & Use Cases

  • Per-subtask logging: Append a concise entry after every subtask with status, evidence, and next steps.
  • Template-backed record-keeping: Create CODEX_RECORD.md from a portable template when missing.
  • Deterministic workflow guidance: Enforces storage rules, environment boundaries, and immediate record updates to maintain a reproducible trail.
  • Use Case: Use during paper/code reproduction tasks, debugging training/evaluation pipelines, and HPC experiments to track progress and evidence.

Quick Start

Run the first subtask and append its entry to CODEX_RECORD.md using the provided script.

Frequently Asked Questions about repro-record-workflow

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate reproducibility logging for HPC experiments?

Track subtask executions in HPC workflows by enforcing deterministic record-keeping rules that append status, evidence, and next steps to CODEX_RECORD.md immediately after every execution.

How do I track debugging steps for training and evaluation pipelines?

Document training pipeline debugging by appending structured entries with status, evidence, and next steps to CODEX_RECORD.md, ensuring a reproducible trail across multiple debugging subtasks.

What is the best way to maintain a reproducibility log for paper reproduction tasks?

The best way to maintain a paper reproduction log is to apply a template-backed record-keeping workflow that creates CODEX_RECORD.md if missing and appends concise per-subtask entries with status and evidence.

Can I enforce environment boundaries and storage rules during experiment record-keeping?

You can enforce environment boundaries and storage rules during experiment record-keeping through deterministic workflow guidance that validates storage paths and immediately updates CODEX_RECORD.md after subtask completion.

Does reproducibility workflow logging require a specific file format?

Reproducibility workflow logging requires a CODEX_RECORD.md file format, which is automatically generated from a portable template if missing, to store structured per-subtask execution logs and evidence.

How do I verify experiment logs generated for HPC workflows?

Verify experiment logs for HPC workflows by running a minimal verify step provided by the Skill's Python scripts, which checks that CODEX_RECORD.md contains guaranteed per-subtask entries with strict evidence logging.