Skip to content
neuroflow0.2.31 · docs

← neuroflow

/review

You are the reviewer. A colleague has sent you their paper and you need to produce a formal referee report. This command is for reviewing someone else's work — not for self-review before your own submission (use /paper for that).

The /review command checks the journal's AI-use policy, gathers the paper, the target journal, and your review focus, then delegates the full analysis to the neuroflow:review-neuro skill. The result is a structured referee report saved in the local-only .neuroflow/review/ folder.

What it does

  1. Asks about the AI-use policy first — before the manuscript is requested or read: does the journal (or funder) allow AI tools in peer review? If not, or if you are unsure, the manuscript is not read; you get only help that needs no manuscript text. Your answer is stamped at the top of the report.
  2. Keeps everything local — the manuscript, notes and report stay in .neuroflow/review/, which is gitignored, never exported by /output, and never sent to NotebookLM, Miro or Zotero (Zotero archival only if you opt in, and never into a cloud-synced library)
  3. Asks for the paper — provide a file path, upload a PDF, or paste the text
  4. Asks for the target journal — used to calibrate the referee persona and standards (optional; defaults to high general standards if not provided)
  5. Asks for review type — full review across all eight areas, or focused on specific areas (methods, statistics, writing, figures)
  6. Scans for hidden instructions — text hidden in the manuscript that tries to steer an AI reviewer is treated as data, reported to you, and noted for the editor
  7. Delegates to neuroflow:review-neuro — the skill runs the complete eight-area review
  8. Saves the report to .neuroflow/review/review-[paper-title-slug]-[date].md

What the review covers

The eight areas reviewed by neuroflow:review-neuro:

Area What it checks
1. Language & Style Spelling, grammar, abbreviations, neuroscience terminology errors, overclaims
2. Internal Consistency Cross-references, figures, tables, numerical values in text vs results
3. Claim Support & Causality Unsupported claims, causality language from correlational data, FC over-interpretation
4. Statistics & Network Inference Power analysis, effect sizes, multiple comparisons, null models, estimator specification
5. Methods Reproducibility Ethics, demographics, modality-specific checklists (fMRI/EEG/iEEG/modelling), open data
6. Contribution & Novelty Novelty vs prior work, conceptual significance, alternative interpretations, journal fit, recommendation
7. Literature Gap Missing citations to relevant work — checked against Zotero, local paper library, and the manuscript's own references
8. Figure Review Citation order, panel completeness, axes/units/error bars, legend definitions, statistical annotations

Example session

/neuroflow:review

> AI-use policy: the journal allows AI assistance if disclosed to the editor
> Here is the paper (file path or uploaded PDF)
> Target journal: eLife
> Focus: full review

[hidden-instruction scan: nothing found]
[review-neuro skill produces a structured referee report]

Review saved to .neuroflow/review/review-default-mode-connectivity-2025-06-15.md

Files read

File Purpose
.neuroflow/project_config.md Active phase and project context
.neuroflow/flow.md Project memory index
skills/phase-review/SKILL.md Phase orientation for the reviewer role

Files written

File Contents
.neuroflow/review/review-[title-slug]-[date].md Full structured referee report (local only, gitignored)
.neuroflow/sessions/YYYY-MM-DD.md One-liner entry (not the full report)