d-laub/dlaub-togo

scientific-writing

Use when writing or continuing a scientific manuscript/paper from a code+results repository — the entry point that routes through fact-finding (story), outlining, and drafting. Use when the user says "write/work on my paper/manuscript".

First seen Jun 17, 2026

Installation

$ npx skills add d-laub/dlaub-togo --skill scientific-writing

Similar popular skills

Related neighbors and high-traction skills in the same topics — useful to compare before installing.

Also in this package

Other skills from d-laub/dlaub-togo · top by installs.

npx skills add d-laub/dlaub-togo

Browse all from d-laub/dlaub-togo

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Not declared
Cursor Not declared
Codex Not declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Not declared

Repository health

Default branch main
Open issues 0
Status Active

Skill metadata

Parsed from SKILL.md frontmatter.

More metadata
type
technique

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 2,984 B
  • docs SUMMARY.md 264 B

History

  1. First seen on skills.sh
  2. First recorded snapshot · 20 installs

SKILL.md

Scientific Writing (orchestrator)

Overview

A three-stage pipeline from a code+results repo to a manuscript draft. Each stage owns a skill and a durable artifact under docs/writing/<story-slug>/ in the consuming repo. This skill routes to the right stage and resumes from whatever artifacts already exist.

Method backbone — the fifteen steps

The pipeline operationalizes Drake & Han's "How to write a scientific paper in fifteen steps" (PLOS Comput Biol, 2025). Its core discipline: finish the thinking before the phrasing, and write only after the research is done. Full detail — every step, with the caption/CARS/connector-test specifics each stage calls on — is in scientific-writing/fifteen-steps.md. Each stage owns a contiguous block:

Steps Stage Delivers
1–2 Story main point + two-part narrative arc
3–12 Outline method/result/finding lists → CARS intro, findings + problem/response discussion, closing thesis, per-paragraph purpose statements, figures
13–15 Draft topic / supporting / concluding sentences

Cross-cutting, and true of this agent too: generative AI is a Steps-13–15 tool (phrasing), not a Steps-1–12 tool (thinking). Do the intellectual work of shaping each stage's artifact before smoothing prose; never let fluency stand in for a claim you haven't backed.

Stages

Stage Skill Reads Writes
1. Story scientific-writing-story the repo story.md
2. Outline scientific-writing-outline story.md outline.md, references.md
3. Draft scientific-writing-draft outline.md, references.md, story.md, style guide draft.md

Routing

  1. Look in docs/writing/*/ for existing artifacts.
  2. Resume at the first incomplete stage:

- no story.md → invoke scientific-writing-story. - story.md but no outline.md → invoke scientific-writing-outline. - outline.md but no draft.md → invoke scientific-writing-draft. - draft.md already exists → all stages complete; ask the user whether to revise draft.md (re-enter scientific-writing-draft) or stop.

  1. If the user names a stage explicitly, honor that.
  2. Confirm the target story dir with the user when more than one exists.

Approval is the user's verbal confirmation in the conversation, not an on-disk marker. Artifact existence drives which stage you resume at; approval gates advancing from one stage to the next (each stage skill ends with "once the user approves, invoke the next stage").

Always announce which stage you're entering and why before invoking it.