enforcing-seed-hygiene

Community

Keep ML runs reproducible, every time.

Authorrocklambros
Version1.0.0
Installs0

System Documentation

What problem does it solve?

This Skill prevents flaky, non-reproducible results in notebooks and ML pipelines by making randomness explicit and consistent across runs, libraries, and platforms.

Core Features & Use Cases

  • Seeds the full stack in one first-cell block, including Python random, NumPy, PyTorch, JAX, TensorFlow, and R.
  • Handles cross-platform sampler reproducibility with CPU and thread pinning when fixed-seed behavior is not enough.
  • Adds repo-level guardrails with a pre-commit check that catches new files missing seed setup.
  • Use it when starting a new notebook, debugging run-to-run drift, preparing reproducible analysis, or standardizing ML project templates.

Quick Start

Use the enforcing-seed-hygiene skill to generate a single first-cell reproducibility setup for my new PyTorch notebook.

Dependency Matrix

Required Modules

None required

Components

references

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: enforcing-seed-hygiene
Download link: https://github.com/rocklambros/rcs/archive/main.zip#enforcing-seed-hygiene

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 536,000+ vetted skills library on demand.