205-datascience-repro

A checklist for making data-science analyses repeatable by recording software versions, random seeds, data sources, settings, and outputs.

In plain words
What is it for?
It helps set up environments, track data lineage, store configuration, save model metadata, log metrics, and use containers.
Why use it?
It reduces the risk that an analysis or model cannot be recreated because its environment, inputs, or processing steps were not recorded.

Cursor rule

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add rules/hamzaamjad/cursor-rules/205-datascience-repro
Clone the repo
git clone --depth 1 https://github.com/hamzaamjad/cursor-rules
Per session 0 Nothing until a file matches its globs; then the whole rule loads.
When invoked 998 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00000 $0.00998
Opus 5 $0.00000 $0.00499
Sonnet 5 $0.00000 $0.00200
Haiku 4.5 $0.00000 $0.00100

Measured 2d ago against content hash 6184d13a66b0, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

205-datascience-repro scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

rules/200-domain/205-datascience-repro.mdc · 89 lines

How it starts

The opening of the file, as written. The whole thing — 89 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Reproducibility Checklist

  • Environment
    • Pin all library versions (requirements.txt, environment.yml, poetry.lock).
    • Record Python version (python --version) and OS info in README.md. Use virtual environments.
  • Randomness control
    • Set seeds for all sources of randomness early in scripts/notebooks (e.g., random.seed(42), numpy.random.seed(42), torch.manual_seed(42)). Document the chosen seed value.
  • Data lineage
    • Record source data locations and checksums (e.g., SHA256) in README or logs.
    • Version control preprocessing scripts. Log script execution commands and arguments.
  • Config
    • Store all hyperparameters and key configuration parameters in separate files (e.g., YAML, JSON, .env) and commit them to version control. Do not hardcode configuration in scripts.
  • Outputs
    • Save model artefacts with metadata (git SHA, run timestamp, config file used).
    • Log performance metrics to an experiment tracker (MLflow, WandB) or a version-controlled CSV/JSON file. Include examples of using MLflow and WandB for tracking.
    • Ensure reproducibility by using Docker or similar containerization tools to encapsulate the environment.
  • Notebooks
    • Clear all cell outputs before committing .ipynb files.
    • Consider exporting finalized notebooks to HTML or PDF for easier review without requiring execution.
  • Ethics & compliance
    • Verify data usage complies with privacy regulations (GDPR, CCPA, etc.).
    • Anonymize or pseudonymize Personally Identifiable Information (PII) where required.
    • Document known biases and limitations of the dataset/model in the README or final report.

Usage: reference @datascience-repro when kicking off an analysis or review.

Validation

  • Check (File Existence & Format): Verify requirements.txt or similar exists and pins versions (e.g., pandas==1.5.3 not pandas>=1.5). Check README.md for env info and seeds. Check for config files. Check for output logs/artifacts.
  • Check (Code Review): Look for hardcoded seeds, paths, or configs. Review data loading and preprocessing steps for lineage tracking. Review notebook diffs for cleared outputs. Check documentation for bias/limitation statements.
  • Check (Tooling): Use experiment tracker UI (MLflow, WandB) to verify runs are logged with parameters and metrics.

Read the full file on GitHub · 89 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 89 lines · 0 tokens per session scan A 6184d13a66b0

Subscribe to this mod's changes

205-datascience-repro is a cursor rule published in the GitHub repository hamzaamjad/cursor-rules (2 stars, last pushed 1y ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 998 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.