Workflow

Workflow

Baca dalam bahasa Indonesia

The six skills answer different questions. The order is not a rigid rule — but skipping it usually means doing the work twice.

   nulis  ──▶  visualisasi-data  ──▶  polish-manuscript  ──▶  submit  ──▶  [editor]  ──▶  revisi
 structure         figures                  prose              gate        decision     response
   ▲                                                                                         │
   └─────────────────── if a reviewer demands the RQ/contribution change ────────────────────┘


   slr-cowork  ──▶  Stage 9 manuscript  ──▶  visualisasi-data  ──▶  polish-manuscript  ──▶  submit

The question each stage answers

Stage	Skill	Its question
Structure	nulis	does each section carry the right moves, and is every RQ traceable from gap to contribution?
Figures	visualisasi-data	does the figure's shape say anything about the thing being studied?
Prose	polish-manuscript	is the prose clear, the argument sound, the claims calibrated?
Gate	submit	will this survive the editor's first ten minutes, or come back before review?
After the decision	revisi	is every reviewer comment answered, and can the editor find each change?
Systematic review	slr-cowork	do the numbers reconcile and will the method hold up?

Why that order

Structure before prose. Polishing sentences in a section whose moves are wrong is work that will be discarded. Fix the frame first.

Figures before prose, too. nulis decides which results deserve to be figures; drawing them then fixes what the Results section actually has to narrate. Writing the prose first and drawing afterwards produces paragraphs that describe a figure nobody has seen yet — and captions written to excuse a figure instead of explaining it.

The gate after prose, before sending. A manuscript with perfect structure and smooth prose is still returned by the editor if the scope missed, an ethics statement is blank, or the word limit is exceeded. None of those three are checked by the two skills before it.

submit stops at a fatal finding. It is a staged gate, not an exhaustive sweep. No point polishing hyphens in a manuscript whose scope has missed — one fatal finding makes the rest irrelevant.

Revision can bounce back to nulis. If a reviewer demands that the research question or the contribution claim change, that is structural work — go back to nulis before the response letter is written.

If only one is installed

Everything still runs. Every SKILL.md has a "Standalone or alongside" section with a table: what deepens if a neighbour is present, and what the fallback is if not.

The rule that holds across all six: a step that cannot run is not counted as passed.

---

← Back · Installation · Requirements · FAQ · Licence