Writing in the Sciences

A Codex skill for drafting and revising scientific, technical, and other evidence-backed writing with a workflow based on Stanford's Coursera course Writing in the Sciences by Dr. Kristin Sainani.

中文说明

Installation

Recommended

Clone this repository into your local Codex skills directory:

mkdir -p ~/.codex/skills
git clone https://github.com/jingkarqi/writing-in-the-sciences-skill.git ~/.codex/skills/writing-in-the-sciences

Manual install or update

If you already have this repository elsewhere, copy the full skill directory into your Codex skills directory:

mkdir -p ~/.codex/skills/writing-in-the-sciences
cp -R SKILL.md agents references assets ~/.codex/skills/writing-in-the-sciences/

The full directory is required because the skill reads structured references from references/, course mirrors from assets/, and agent metadata from agents/openai.yaml.

Usage

Invoke the skill directly in Codex:

$writing-in-the-sciences

Revise this abstract for a clinical research audience. Keep the claims within the supplied evidence and tighten the prose.

Or ask for it explicitly in natural language:

Use $writing-in-the-sciences to rewrite this introduction for a review article. My audience is graduate students, and the main take-home message is that resistance evolves faster than policy.

For best results, provide:

• source text, notes, reviewer comments, or bullet-point facts
• target genre and audience
• the main take-home message
• length or section constraints
• any figures, tables, or screenshots whose wording must be preserved

The skill is designed to surface missing evidence instead of inventing it.

Overview

This repository packages a reusable writing workflow derived from the Stanford course and organizes it as a Codex skill. It is intended for scientific or evidence-backed writing tasks, including:

• manuscript sections and abstracts
• review articles
• grants and specific aims
• recommendation letters
• research-oriented personal statements
• peer review and reviewer responses
• lay summaries and science-news style explanations
• general revision of scientific or technical prose

Workflow

The skill follows the same broad sequence for drafting and revision:

1. Classify the task by genre so the right reference material is loaded.
2. Confirm the available inputs, audience, constraints, and take-home message.
3. Do pre-writing first: organize facts, claims, evidence, and gaps before prose.
4. Draft for structure and paragraph purpose before polishing sentences.
5. Revise in passes for clutter, verbs, flow, jargon, and overclaiming.
6. Run final checks for consistency, evidence alignment, and numerical accuracy.

At every stage, the skill prioritizes clarity over flourish and keeps claims bounded by the supplied source material.

Repository Layout

• SKILL.md: the skill definition and operating rules
• agents/openai.yaml: display metadata and the default agent prompt
• references/: concise reference sheets for style, process, manuscripts, grants, peer review, broader audiences, and more
• assets/: mirrored course notes and supporting images used when the summaries are not enough

Sources

The workflow and reference material in this repository are based on:

• Stanford University Coursera course: Writing in the Sciences
• Dr. Kristin Sainani's course structure and writing guidance
• local summaries and mirrored source notes included under references/ and assets/

This repository is a community-packaged skill and is not an official Stanford or Coursera product.

License

This repository is licensed under the MIT License. See LICENSE.