Paper workflow

Paper Workflow

Overview

Use this skill as the routing map for paper work. It does not replace the specialized skills. It tells you which one to use next.

Default assumption: unless the user explicitly names a conference venue, the manuscript should follow the journal-oriented, Nature-style workflow.

When writing follows experiments or analysis performed across multiple sessions, freeze a compact handoff first: current paper story, supported findings, unresolved decisions, and the figure list that actually carries those findings.

Routing Map

Use:

• paper-bootstrap when a project is new, messy, or missing a source of truth
• nature-portfolio-playbook when the target is Nature, Nature Methods, Nature Biotechnology, or the venue fit among those journals is still unclear
• scientific-writing when drafting or rewriting sections in prose
• manuscript-optimizer when the paper's claim structure, evidence chain, terminology, or prose need revision
• results-section-revision when the Results section is scientifically stable but the subsection flow, paragraph openings, or argument progression still feel jumpy
• figure-planner when the main bottleneck is figure logic, panel roles, or legend sync
• citation-verifier when bibliography hygiene or source verification is the bottleneck
• submission-audit when the paper is near submission or resubmission and needs a preflight pass
• rebuttal-response when reviewer comments exist and a response letter plus aligned manuscript edits are needed
• conference-paper-writing only when the user explicitly wants a conference paper for venues such as NeurIPS, ICML, ICLR, ACL, AAAI, or COLM

Default Sequence

For most papers under /data/boom/Papers, prefer this order:

1. paper-bootstrap
2. nature-portfolio-playbook when venue fit or article type is uncertain
3. Refresh notes/project_truth.md, notes/result_summary.md, and notes/paper_handoff.md after any major experimental or figure update
4. scientific-writing or manuscript-optimizer
5. figure-planner
6. results-section-revision when Results is substantively right but narratively abrupt
7. citation-verifier
8. submission-audit
9. rebuttal-response after external review

Common Mistakes

• using conference-style writing skills by default for journal manuscripts
• polishing sections before the active manuscript source of truth is clear
• rewriting the manuscript from experiment memory instead of a current result_summary.md or paper_handoff.md
• editing figure legends late without rechecking the Results text
• postponing Nature Portfolio article-type or venue-fit decisions until after the paper is mostly rewritten
• treating citation formatting as the same thing as citation verification
• writing a response letter before deciding the underlying manuscript edits