Scientific Workflow Tools

Use this skill when the user needs higher-level research method support rather than raw database lookup or computation.

Typical triggers:

• generate or compare mechanistic hypotheses from observations
• review a manuscript draft for rigor, missing controls, or overstated claims
• build a reproducibility checklist before submission or release
• structure a scientific report, review, or response-to-reviewers plan
• identify missing controls, statistical gaps, or reporting-standard issues

Bundled Asset

• templates/reproducibility_checklist.py

Preferred Workflow

1. Restate the research question, claim, or draft under review.
2. Separate what is observed from what is inferred.
3. Enumerate methodological risks before proposing fixes.
4. Use the checklist template to create a durable artifact for reporting or project tracking.
5. Keep outputs explicitly on the research side. Do not cross into patient-level diagnosis or treatment planning.

Reproducibility Checklist

python3 templates/reproducibility_checklist.py \
  --profile omics \
  --output research/omics_checklist.md \
  --summary research/omics_checklist.json

Supported baseline profiles:

• general
• omics
• ml
• clinical-research

Use the generated checklist as a starting artifact, then tailor it to the exact study.

Working Rules

• Hypotheses should be testable and distinguish observation from mechanism.
• Peer-review style critique should prioritize reproducibility, controls, statistics, and claim scope.
• Scientific writing support should strengthen structure and rigor, not fabricate citations or results.
• Reporting-guideline and checklist outputs are planning artifacts, not proof that the study is compliant.

Related Skills

For literature search outputs and evidence tables, activate literature-review-tools. For clinical-study design and reporting-guideline selection, activate clinical-research-tools. For numerical statistical execution, activate stat-modeling-tools or survival-analysis-tools. For experiment suggestion or bounded closed-loop optimization, activate bayesian-optimization-tools. For figure generation, activate scientific-visualization-tools. For bioinformatics, chemistry, or docking execution, activate the corresponding domain skill instead of keeping the task abstract.