Guide · Agent skill
How to use the nature-statistics agent skill for manuscripts
Updated · written with AI from the project's own documentation and checked against our catalogue data
The nature-statistics skill is an add-on for an AI agent that reviews and rewrites the statistical reporting in a scientific manuscript. It checks what the independent unit of analysis is, whether tests, sample sizes, effect sizes and uncertainty are stated, and whether figure legends match the statistics. This guide explains what the skill does, how to install it, how to ask for it in plain language, and where its limits are.
Nature Statistics Reporting ReviewYuan1z0825/nature-skills · Apache-2.0 Download the skill Quick start
- 1Download the ZIP. Get the nature-statistics skill ZIP from this page.
- 2Unzip into a skills folder. Extract it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder for one project.
- 3Or upload it in the Claude apps. In the Claude apps, open Settings, then Capabilities, then Skills, and upload the ZIP.
- 4Paste your manuscript text. Give the agent your Statistical analysis or Methods subsection, Results paragraphs, figure legends or reviewer comments, with your sample sizes and replicate levels.
- 5Ask for the task you need. Say whether you want an audit, a rewrite, a draft, a reviewer reply or a figure-legend check, and review the AUTHOR_INPUT_NEEDED questions before using the text.
What is an agent skill and what does this one do?
An agent skill is a folder of instructions and reference files that an AI agent loads when a task matches its description. This skill, named nature-statistics, is built for reporting and review of manuscript statistics, mainly for Nature-style and other high-impact journal submissions.
Its default stance is to make the design visible. It separates what was measured, what unit was analysed and what inference was claimed. It treats the independent experimental unit as the default n, so cells, fields of view, repeated readings and technical replicates are not counted as independent samples unless the hierarchy supports it.
- Audit a Statistical analysis or Methods section for completeness.
- Rewrite statistical statements in Results, figure legends and source-data notes.
- Flag pseudoreplication, nested data, multiple comparisons, interaction claims and significance-only wording.
- Draft conservative replacement wording and point-by-point replies to reviewer statistics comments.
- Mark anything missing as AUTHOR_INPUT_NEEDED instead of guessing.
Install the nature-statistics skill
Download the ZIP of the skill from this page. Unzip it into your personal skills folder ~/.claude/skills/, or into a project's .claude/skills/ folder if you want the skill to apply only to that project.
If you use the Claude apps instead, upload the ZIP under Settings, then Capabilities, then Skills. The skill is licensed under Apache-2.0.
- Personal install: ~/.claude/skills/
- Project install: the .claude/skills/ folder inside your project
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
Check the requirements before you start
The skill lists no required software or tools. It does not run scripts and does not use the network, so nothing is sent to an external service by the skill itself. Whether your agent sends your text to its own model provider depends on that agent, not on this skill.
The skill works from the text you give it. Paste the Statistical analysis or Methods subsection, relevant Results paragraphs, figure legends, reviewer comments or a table of comparisons. Include sample sizes, replicate levels, the experimental unit, tests and multiple-comparison methods where you have them.
- No listed dependencies
- No scripts are run
- No network use
- Input can be partial; the skill then says which parts it could not assess
Trigger the skill with natural requests
You do not need a special command. Ask in plain language, and mention the task. The skill's description includes Chinese and English triggers such as 统计审查 (statistical review), 统计方法小节 (statistical methods subsection), 图注统计 (statistics in figure legends) and reviewer concerns.
Be specific about what you want back: an audit, a rewrite, a draft, a reviewer reply or a figure-legend check. The skill first classifies the task, then extracts the design before judging the claims.
- Check whether this Statistical analysis section is complete for a Nature-style submission.
- Are the n, error bars and p values in this figure legend stated clearly?
- A reviewer says the statistics are insufficient. Draft a point-by-point response with a fix list.
- Rewrite this Results paragraph so it does not overstate the effect.
Read the output and act on it
For a full audit, the skill returns a scope section, a list of major issues labelled P0, P1 or P2 with evidence and fixes, ready-to-paste revised text, an AUTHOR_INPUT_NEEDED list of short factual questions, and a reviewer-risk note. For a clean drafting request with enough detail, it returns a shorter draft with reporting notes on n, tests, multiple comparisons, software and unresolved fields.
Treat the AUTHOR_INPUT_NEEDED items as questions for you to answer from your records. The skill is designed not to invent sample sizes, tests, software versions, exclusion rules or corrections, so those gaps stay visible until you fill them.
- P0: likely to undermine the result, such as pseudoreplication.
- P1: important reporting or interpretation risk.
- P2: clarity and presentation risk.
- Check every rewritten sentence against your own data before submitting.
Ask for new analyses only with raw data
By default the skill reviews and rewrites reporting. It recomputes analyses only when you supply raw or summary data and explicitly ask for a reanalysis or figure-statistics check. Without data, it cannot confirm that a test was run correctly.
The skill also states that it is not a substitute for a statistician reanalysing raw data. For clinical or regulatory trial decisions, it limits itself to reporting checks and does not give medical statistical advice beyond that.
- Without data: review and wording only.
- With data and an explicit request: concrete reanalysis or figure-statistics check.
- Clinical or regulatory decisions: outside the skill's scope.
What people use it for
- Pre-submission check of a Statistical analysis section for a Nature-style paper.
- Rewriting figure legends so n, error bars and tests are stated clearly.
- Drafting a conservative, point-by-point reply to a reviewer who questions the statistics.
- Teaching students or lab members what a statistically transparent paper should report.
Frequently asked questions
How do I install the nature-statistics skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. In the Claude apps, you can instead upload the ZIP under Settings, then Capabilities, then Skills.
Is the nature-statistics skill free?
The skill is licensed under Apache-2.0, as listed on this page. Check the licence file in the download for the full terms.
What do I need to run nature-statistics?
The skill lists no required dependencies. It does not run scripts and does not use the network. You need an agent that supports skills and the manuscript text you want reviewed.
Can nature-statistics recalculate my p values?
Only if you supply the raw or summary data and explicitly ask for a reanalysis. Otherwise it reviews and rewrites the reporting and does not invent p values or sample sizes.
Is nature-statistics a replacement for a statistician?
No. The skill describes itself as a reporting and review aid. It is not a substitute for a statistician reanalysing raw data, and it does not give clinical or regulatory trial statistical advice beyond reporting checks.
What should I use instead if I need full statistical analysis?
For reanalysis or study design decisions, work with a statistician who can see your raw data and protocol. This skill is best used to check and tighten how your results are reported.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use nature-statistics if you are preparing a Statistical analysis or Methods section, answering reviewer statistics concerns, or learning what a transparent paper should report. Pick a statistician instead if you need reanalysis of raw data, or clinical or regulatory trial advice, since the skill explicitly limits itself to reporting checks.
Download the skill