reproducible-analysis

What problem does it solve?

Reproducible analysis is hard when steps are scattered across scripts and outputs; this skill provides a unified, machine-readable journal to capture every data operation so analyses can be replayed and verified end-to-end.

Core Features & Use Cases

• Automatic generation of a <analysis-name>.journal.jsonl file that records each operation as a JSON object with fields like seq, ts, op, tool, input, input_sha256, output, output_sha256, duration_ms, and note.
• Supports deterministic replay across environments, enabling auditors and teammates to reproduce results given the same inputs and toolchain.
• Works with common data formats (CSV, JSON, Parquet) and multiple data-processing steps (indexing, filtering, joining, statistics).

Quick Start

Create the first reproducible journal entry by running an analysis step and appending its details to a file named <analysis-name>.journal.jsonl.