sci-workflow

Science Workflow Router (sci-workflow)

Overview

This is the router. It does not replace any specialized skill. It tells you **which sci- skill to use at the current stage* of a manuscript aimed at Science (AAAS).

Default assumption: unless the user states otherwise, the target is Science (the flagship research journal), not Science Advances, Science Translational Medicine, or other family titles. Family titles share house style but differ on scope and length — flag the difference if the user names one.

When to trigger

• "What should I do next with this draft?"
• A draft arrives and you must diagnose the current bottleneck.
• The user is switching between experiments, writing, and revision and has lost the thread.
• Reviews arrive from Science and you need to switch into rebuttal mode.

The single most important gate

Science rejects most submissions without external review. The editorial bar is broad significance and general interest, not technical correctness alone. So the first question is never "is the science right?" — it is "does this clear the desk-reject filter for a general audience?" Route to sci-fit first, always.

Routing table

| Current symptom | Next skill |

|-------------------------------------------------------------|---------------------|

| Not sure the result is broad/general-interest enough | sci-fit |

| Result is significant but the "why it matters" is buried | sci-framing |

| No one-sentence summary; abstract reads narrow/jargon-heavy | sci-abstract |

| Unsure whether this is a Research Article or a Report; over length | sci-writing |

| Figures over panel/word budget; fonts/colors non-compliant | sci-figures |

| Stats under-reported; n/error bars/tests unclear; not reproducible | sci-statistics |

| No data/code/materials availability plan | sci-data |

| References not in Science numbered style | sci-citation |

| Need an editor-facing pitch for why Science should review it | sci-cover-letter |

| About to submit; need a preflight checklist | sci-submission |

| Received reviews / an R&R decision | sci-rebuttal |

Default order

1. sci-fit — clear the broad-significance / general-interest bar first
2. sci-framing — lock the conceptual advance and the "why now"
3. sci-writing — choose format (Research Article vs Report) and hold length
4. sci-figures — finalize display items within budget
5. sci-statistics — rigor & reproducibility reporting
6. sci-data — data / code / materials availability
7. sci-abstract — abstract + one-sentence summary (late polish)
8. sci-citation — reference style pass
9. sci-cover-letter — significance-forward editor pitch
10. sci-submission — preflight
11. sci-rebuttal — after review

> sci-abstract and sci-citation are late-stage polish. Do not perfect the abstract before the significance and format are settled.

Decision shortcuts

• "A specialist would care, but would a geneticist and a physicist?" → sci-fit
• "My intro starts with background, not with the advance" → sci-framing
• "My Report main text is 4,000 words" → sci-writing (Reports target ~2,500)
• "Methods are in the main text" → sci-writing (Reports push Methods to Supplementary)
• "Bar chart with no individual data points / no n" → sci-statistics + sci-figures
• "No accession numbers / no code repo" → sci-data
• "Editor needs to see the hook in 3 sentences" → sci-cover-letter

How Science differs from its siblings and from Nature

• Science (Reports / Research Articles) vs Science Advances (open-access, broader acceptance, longer): if the work is solid but not top-1% general interest, Advances may be the realistic target — say so.
• Science vs Nature: both gate on broad significance, but Science requires a one-sentence summary and uses a single numbered reference list with full author names; Nature uses a different reference and summary-paragraph convention. Do not copy a Nature-formatted manuscript over without a style pass (sci-citation, sci-abstract).

Stage ledger (copy into the conversation and keep it updated)

Track the manuscript's position in the pipeline with this ledger so every routing decision is explicit and auditable:

SCIENCE (AAAS) MANUSCRIPT STAGE LEDGER
======================================
Target venue      : Science | Science Advances | reroute → ______
Format            : Report | Research Article | undecided
Stage gates (mark ✓ / ✗ / n/a, with date):
  [ ] sci-fit          significance rung ≥3, general-interest test passed
  [ ] sci-framing      one-sentence advance locked; why-now named
  [ ] sci-writing      format chosen; main text within word budget
  [ ] sci-figures      display items ≤ budget; fonts/colors compliant
  [ ] sci-statistics   n, tests, effect sizes, error bars reported
  [ ] sci-data         deposition + accession numbers + availability statement
  [ ] sci-abstract     ≤125-word abstract + one-sentence summary
  [ ] sci-citation     numbered refs, order of appearance, full author lists
  [ ] sci-cover-letter significance-forward editor pitch drafted
  [ ] sci-submission   preflight checklist clean
  [ ] sci-rebuttal     (post-review only) point-by-point response
Current bottleneck : ______            Next skill: sci-______
Open risks         : over-claiming? length? missing deposition? ______

Re-emit the ledger after each sub-skill completes; a gate marked ✗ upstream (especially sci-fit) invalidates every ✓ downstream of it.

Anti-patterns

• Do not skip sci-fit and start polishing prose — the modal outcome is desk rejection.
• Do not let sci-figures beautify panels before the format/length decision in sci-writing.
• Do not generate a rebuttal before the main text is actually revised.