GuoZheng-de-BaBa/nursing-research-skills

What it is

Nursing Research Skills packages nursing research workflows as 18 AI skills, each a self-contained directory with a SKILL.md, references and templates. A host AI reads a skill and carries out the work; the repository supplies the method, checklists and output templates, not a standalone research system.

The coverage follows a research project's stages: literature search and reference checking (search skill, reference check), paper reading and evidence cards (paper reader, evidence check), study design and data stewardship (design, data), local statistics (statistics), manuscript writing variants (cross-sectional, review, general manuscript), polishing, scientific and mechanism figures, presentations, simulated peer review and response letters. Version v0.4.0 adds resources for two parallel RCTs, with a reproducible synthetic example (RCT example).

It targets nursing researchers who already work with an AI host that can read local files or a skills directory.

Quick start

The README recommends trying a skill on synthetic material before using real data. Prepare a git client (or download a Release archive and verify its SHA-256) and an AI host able to read the skill file.

git clone https://github.com/GuoZheng-de-BaBa/nursing-research-skills.git
cd nursing-research-skills

Then point the AI at a skill and an example input, as described in the quick reference:

Please read skills/nursing-data-stewardship/SKILL.md and use the synthetic material in examples/v0.2/nursing-data-stewardship/input.md.
Deliver the data-readiness conclusion, the evidence that still needs supplying, and the work that can continue now.

The expected output is a structured data-readiness review rather than a finished paper. For a fuller pipeline, the statistics synthetic demo shows an R workflow whose result ledger can be passed to the writing or figure skills.

Requirements and constraints

• Python helper scripts require Python 3.10+; PDF extraction needs requirements.txt, and the validation tools use requirements-dev.txt. R teaching scripts assume a standard R installation.
• Without a script environment, the text and blank templates are still usable.
• The public package contains no private papers, lab materials or real research data; only currently authorized material is read. Privacy documentation warns that source.json keeps original filenames and extracted full text, and that drafts, logs, CSVs and figures are not automatically de-identified (privacy notes).
• The repository states it is not a clinically validated automatic research system and does not replace the author, statistician or research team.
• The suite's own validation separates program tests, independent AI behavior evaluation, figure-format checks and unverified areas; passing structural checks does not prove scientific judgment is correct.

Maintenance and license

The current version is v0.4.0, with the 18 skill entry points kept and older skill IDs and JSON contracts preserved; a mapping document records where earlier capabilities were merged. Releases are distributed through GitHub Releases. Original code, templates and synthetic cases are MIT-licensed, with third-party method attributions listed in THIRD_PARTY_NOTICES. Contribution requests ask for minimal synthetic cases and warn against uploading private material (contributing).

HotRepo's take

HotRepo's only recorded snapshot for this project is the 2026-11-02 new-project list (Python): 120 stars, 17 gained in a roughly two-hour window, no recorded best GitHub daily rank. That is a short observation of a newly created repository, not evidence of sustained adoption or of the quality of the research guidance. The more useful signal for readers is the repository's own candor: it separates executable test results from AI behavior evaluation and from clinical validity, and it explicitly excludes private data from the public package.