choosing-research-skillslisted

# Choosing Research Skills ## Overview There are dozens of published research and paper-writing skill packs, several with tens of thousands of stars. They vary enormously in what they actually ship, and **star count does not track usefulness** — several of the most starred carry community-growth material, and one widely installed citation checker contains no code at all. This skill routes a researcher to the right pack for their task and field, and installs it with a verified command. **It contains none of their content and replaces none of them.** ## The filter that matters Measured on this ecosystem: a modern model already writes a competent Related Work section, already gives a rigorous reviewer critique, and already catches fabricated references when asked directly. Packs whose substance is prose telling the model *how to write* therefore add little. **Prefer packs that ship executable tooling** — scripts, database queries, validators, gates. That is the axis along which these packs actually differ. ## Route by task | Task | Pack | Why this one | |---|---|---| | Bioinformatics and computational biology: sequence alignment, structure prediction, single-cell, pathways | `K-Dense-AI/scientific-agent-skills` | 148 skills over ~78 databases: AlphaFold, BioPython, BLAST, scanpy/anndata, cellxgene, ESM, cobrapy. Real tooling, not writing advice. | | Figures for a submission; Nature-family house style; statistics reporting | `Yuan1z0825/nature-skills` | `validate_figure.p