Traction scoreGitHub stars can be faked, so popularity alone can be misleading. Traction Score looks for broader signs of recent attention, adoption, and active maintenance.
This collection provides specialized utilities for academic researchers conducting literature reviews and evidence synthesis. It streamlines the process of searching scientific papers and retrieving granular experimental data—such as sample sizes, quality scores, and methodology—that often remains hidden in full-text studies.
WHO IT'S FOR
Academic researchers
conducting literature reviews and evidence synthesis
Life sciences researchers
searching bioRxiv preprints for latest findings
Scientific writers
generating and validating BibTeX citations
Bioinformatics and drug discovery specialists
extracting structured experimental data from studies
Compatible AgentsThe repository documents support for these agents. The skills may also work with other agents that can load SKILL.md files, but they may need some setup or small changes.
Follow the documented setup, then try a first task.
I’m using
Install the complete collection globally for Claude Code
Installs: Complete Qinyan Academic Skills collection; the listed paths are a bounded selection from the documented five-skill Nature-style suite. · Claude Code
Before you start
Bash
Git
curl
Platform: On Windows, run the installer in WSL or Git Bash.
Run this command in a terminal. The default destination is Claude Code's global skills directory.
Install the complete collection globally for Codex
Installs: Complete Qinyan Academic Skills collection targeted to Codex; the listed paths are a bounded selection from the documented five-skill Nature-style suite. · Codex
Before you start
Bash
Git
curl
Platform: On Windows, run the installer in WSL or Git Bash.
In a terminal, define the installer URL. Keep this variable in the same terminal session for the next step.
Install qinyan-nature-writing globally for Claude Code
Installs: Single skill: qinyan-nature-writing. · Claude Code
Before you start
Bash
Git
curl
Platform: On Windows, run the installer in WSL or Git Bash.
Run this single-skill installation command in a terminal. Without a project or tool flag, the documented default is Claude Code's global skills directory.
In that same terminal session, install the documented scanpy skill. The absent tool and project flags retain the documented global Claude Code default.
Install scientific-writing into the current Claude Code project
Installs: Single skill: scientific-writing, installed at project scope. · Claude Code
Before you start
Bash
Git
curl
Platform: On Windows, run the installer in WSL or Git Bash.
Run from: Current project directory
Open a terminal in the project that should receive the skill, then define the installer URL. Keep this variable in the same terminal session for the next step.
Build a claim–evidence outline for one manuscript section
Uses qinyan-nature-writing
The target manuscript section
Verified results, figures, tables, experimental notes, or evidence pointers
The intended audience and target journal, if known
The desired word limit or delivery format
Use the qinyan-nature-writing skill on the supplied manuscript materials. For one specified section, produce only a concise argument spine, a claim–evidence map, and a three-paragraph outline. Identify unsupported claims or missing facts as AUTHOR_INPUT_NEEDED; do not invent results, references, mechanisms, statistics, or journal policies.
Polish one academic paragraph with fidelity checks
Uses qinyan-nature-polishing
One existing Chinese or English academic paragraph
Its manuscript section and intended audience
Target journal and word limit, if applicable
Any facts, terminology, numbers, or citation meanings that must remain unchanged
Use the qinyan-nature-polishing skill in its default annotated mode to polish the supplied paragraph. Return the revised paragraph, the key structural or language edits, scientific-meaning risks, and any AUTHOR_INPUT_NEEDED items. Preserve every number, unit, comparison direction, citation intention, and scientific qualification; do not strengthen correlation into causation or a trend into significance.
Review one manuscript subsection for its highest-priority concerns
Uses qinyan-nature-review
The manuscript subsection as text or Markdown
Any figures, tables, data, or methods referenced by that subsection
The manuscript's core claim and intended audience, if known
A clear note identifying any materials that were not supplied
Use the qinyan-nature-review skill for a focused pre-submission review of the supplied subsection. State the input boundary, then report no more than three highest-priority concerns. For each concern, include a stable Concern ID, Issue key, severity, claim pointer, evidence pointer, why it matters, and a verifiable resolution test. Mark unavailable evidence as NOT_LOCATABLE and do not predict acceptance or invent reviewer identities, experiments, sources, or manuscript details.
Claude
Codex
Cursor
Gemini CLI
OpenClaw
Amp
OpenCode
DeepSeek