ai-research-reproduction
The ai-research-reproduction skill provides a structured workflow for reproducing deep learning research repositories. It guides agents through intake, environment setup, execution, and reporting while enforcing strict documentation of deviations and evidence.
Is ai-research-reproduction safe to install?
Review before installing: our audit of ai-research-reproduction's source files found 0 shell commands, 0 external URLs, file reads and writes (medium risk). Every command and URL listed appears verbatim in the skill's source. The skill writes files to the repro_outputs/ directory and may modify repository files if patches are required for reproduction.
How we audit skills: our security review methodology.
Who is this skill for?
Researchers and developers who need to reproduce results from AI code repositories using a minimal-trustworthy approach.
What can you do with it?
- Documented inference reproduction
- Documented evaluation reproduction
- Documented training startup verification
- Repository-grounded experiment analysis
- Standardized reproduction reporting
How good is this skill?
Quality score: 5/10. The skill documentation is clear, follows a logical workflow, and provides specific instructions for maintaining scientific integrity during reproduction.
What does the skill file contain?
# ai-research-reproduction ## Purpose Use this as the Rigor Reproduce compatible skill slug for README-first deep learning repository reproduction. The installed slug remains `ai-research-reproduction` for compatibility. The skill guides the agent toward a minimal trustworthy run with auditable evidence; it should not micromanage implementation details that the model can infer from the repository. Reproduction is not "make it run by changing anything"; it means faithfully reading the README, environment, weights, datasets, and documented commands, then recording results and deviations. Star...
Frequently asked questions
What is the primary goal of this skill?
The skill aims to faithfully reproduce results from a repository by reading the README, environment, weights, and datasets, then recording all results and deviations in a standardized format.
Does this skill allow code changes?
The skill prefers no repository edits. If edits are necessary, it requires them to be conservative, auditable, and documented in a specific branch and a PATCHES.md file.
What outputs does the skill produce?
It produces a standardized bundle in the repro_outputs/ directory, including SUMMARY.md, COMMANDS.md, LOG.md, SCIENTIFIC_CHANGELOG.md, COMPARABILITY_REPORT.md, status.json, and optionally PATCHES.md.
Related skills
web-design-guidelines
440.2KDevelopers and designers who need to verify UI code compliance with interface standards
The web-design-guidelines skill audits UI code against Web Interface Guidelines. It fetches current rules from a remote source and compares them against specified local files.
paper-context-resolver
140.4KDevelopers and researchers performing deep learning repository reproduction who encounter specific technical ambiguities in documentation
The paper-context-resolver skill provides targeted information from research papers to resolve specific gaps in deep learning repository reproduction. It focuses on technical details like dataset splits, preprocessing steps, evaluation protocols, checkpoint mappings, and runtime assumptions.
repo-intake-and-plan
139.8KAI agents performing repository-based research and reproduction tasks
The repo-intake-and-plan skill scans repositories to identify project structure and documented commands. It classifies tasks into inference, evaluation, and training categories to generate a reproduction plan for an orchestrator.
env-and-assets-bootstrap
139.8KUsers performing deep learning repository reproduction who require environment and asset path configuration before executing code
The env-and-assets-bootstrap skill prepares conservative conda environments, checkpoint paths, and dataset locations for deep learning repository reproduction tasks.