Cross platform reproducibility configuration

Skill HolobiomicsLab/asb-skill-collections/collections/metabolomics/v2/skills/cross-platform-reproducibility-configuration

Curated, evidence-grounded skill and software-tool collections for scientific AI agents, generated by the AgenticScienceBuilder

Install

npx -y skills add HolobiomicsLab/asb-skill-collections --skill cross-platform-reproducibility-configuration

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

14 stars14 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when when you have documented pinned package versions for a Python-based scientific implementation (e.g., from a README or project documentation) and need to distribute a reproducible computational environment to other users or CI/CD systems.

The file declares its own license as CC-BY-4.0. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

7.8 KB, ~1.5k tokens by cl100k_base, as published. Nobody here has run it

cross-platform-reproducibility-configuration

License: restricted — no clear open-source license detected for the underlying tool; verify licensing before commercial use or redistribution.

Summary

Generate and validate pinned software dependency specifications in multiple formats (requirements.txt and environment.yml) to enable exact reproduction of a computational environment across platforms. This skill ensures that all transitive dependencies and their exact versions are captured and compatible with the target Python runtime.

When to use

When you have documented pinned package versions for a Python-based scientific implementation (e.g., from a README or project documentation) and need to distribute a reproducible computational environment to other users or CI/CD systems. Apply this skill when downstream users must achieve bit-identical or functionally identical results without dependency resolution ambiguity.

When NOT to use

When dependencies are specified only by major version or version ranges without exact pinning — the skill requires concrete version numbers, not semantic ranges like '>=1.0,<2.0'.
When the target runtime environment is not Python or uses a non-standard package manager (e.g., R-only projects, compiled binaries without package metadata).
When the project has no documented version pinning and dependency versions must be inferred or reverse-engineered from source code — this skill assumes authoritative source documentation exists.

Inputs

Project README or documentation containing pinned software versions
Target Python version specification
List of direct dependencies with exact version numbers

Outputs

requirements.txt (pip-compatible dependency specification with pinned versions)
environment.yml (Conda-compatible environment specification)
Dependency compatibility validation report

How to apply

Parse the project documentation (README, setup.py, or pinned version lists) to extract all direct dependencies with their exact version numbers. Generate a requirements.txt file in the format 'package==version' for each pinned dependency. Generate a parallel environment.yml (Conda) specification listing the same versions. Validate syntax of both files and use a dependency resolver (e.g., pip-tools or Conda's solver) to confirm that all listed versions are compatible with the target Python runtime (e.g., Python 3.6.12 in the massNet case). Export both formats to enable reproducibility across different installation workflows (pip on Linux/macOS/Windows and Conda on any platform).

Related tools

Python (Runtime environment for which dependency specifications are generated and validated)
pip (Package installer and requirements.txt format validator)
Conda (Cross-platform package manager for which environment.yml specifications are generated and validated)
pip-tools (Dependency resolver and requirements.txt syntax validator)
Keras (Deep learning framework with pinned version requirement (2.2.0 with Tensorflow 1.8.0 backend)) — github.com/wabdelmoula/massNet
Tensorflow (Backend for Keras with pinned version requirement (1.8.0)) — github.com/wabdelmoula/massNet

Examples

pip install -r requirements.txt; conda env create -f environment.yml

Evaluation signals

requirements.txt contains all nine pinned dependencies in format 'package==version' with no version ranges or wildcards
environment.yml syntax is valid YAML and specifies dependencies in Conda format with matching pinned versions
Dependency resolver confirms all pinned versions are mutually compatible and compatible with Python 3.6.12
Both requirements.txt and environment.yml can be used to successfully install the environment without dependency conflicts or unmet version constraints
Checksum or hash verification of installed packages matches expected versions across independent installation runs

Limitations

Version pinning does not account for binary incompatibilities or platform-specific wheels (e.g., numpy 1.15.4 may not have pre-built wheels for all platforms); validation must be run on target platforms.
Requirements.txt format does not capture environment variables, system dependencies, or CUDA/GPU library versions required by Tensorflow 1.8.0; supplementary documentation is needed.
Very old pinned versions (e.g., Python 3.6.12, Tensorflow 1.8.0) may have security vulnerabilities or lack compatibility with modern operating systems; reproducibility may be achievable only in containers or legacy environments.
Transitive dependencies of pinned packages are not explicitly listed; if a pinned package has its own dependencies with floating versions, reproducibility across time may degrade as those transitive packages are updated.

Evidence

[other] The massNet implementation requires Python 3.6.12, Keras 2.2.0 with Tensorflow 1.8.0 backend, and eight pinned dependency packages: numpy 1.15.4, sklearn 0.23.2, scipy 1.0.0, seaborn 0.9.0, Pandas 1.1.1, and h5py 2.7.1.: "Python 3.6.12, Keras 2.2.0 with Tensorflow 1.8.0 backend, and eight pinned dependency packages: numpy 1.15.4, sklearn 0.23.2, scipy 1.0.0, seaborn 0.9.0, Pandas 1.1.1, and h5py 2.7.1"
[other] Generate a requirements.txt file listing each dependency with its pinned version in the format 'package==version'.: "Generate a requirements.txt file listing each dependency with its pinned version in the format 'package==version'"
[other] Validate the requirements.txt syntax and confirm all listed versions are compatible with Python 3.6.12 using dependency resolver tools.: "Validate the requirements.txt syntax and confirm all listed versions are compatible with Python 3.6.12 using dependency resolver tools"
[other] Export environment specification in both requirements.txt and environment.yml (Conda) formats for reproducibility across platforms.: "Export environment specification in both requirements.txt and environment.yml (Conda) formats for reproducibility across platforms"
[readme] We have implemented our machine learning model using the following software items: 1- Python(3.6.12) 2- Keras (2.2.0) with a Tensorflow(1.8.0) backend. 3- Packages: numpy(1.15.4), sklearn(0.23.2), scipy(1.0.0), seaborn (0.9.0), Pandas(1.1.1.), and h5py(2.7.1): "Python(3.6.12) 2- Keras (2.2.0) with a Tensorflow(1.8.0) backend. 3- Packages: numpy(1.15.4), sklearn(0.23.2), scipy(1.0.0), seaborn (0.9.0), Pandas(1.1.1.), and h5py(2.7.1)"