researchers-review

/researchers-review [--personas ]

Peer-review an existing study, claim, or research artifact through multiple researcher lenses.

⚠️ NOT ACADEMIC ADVICE. This review is a reasoning exercise from three persona lenses, not real peer review.

What this does

Dispatches researcher personas to peer-review the artifact at <path>. Default panel runs three personas in parallel (Gould, Sagan, Wilson — synthesis-strong), each with a different lens. Output is a consolidated review with per-persona verdicts and a single highest-leverage recommendation.

When to use

• A draft study needs scrutiny before being shared.
• An existing claim or finding benefits from a multi-lens review.
• A literature review feels off and needs named diagnoses.
• A research artifact needs synthesis-and-skepticism check.

Not for: real journal peer review (this skill is craft, not actual double-blind review); pixel-level citation nits; debate format (filed for v1.0 as /researchers-debate).

Instructions for Claude

When this skill is invoked with a <path> argument and optional --personas:

1. Resolve the project root. Verify CLAUDE.md exists.

2. Read the project specification:

   • CLAUDE.md, README.md, prior studies at research/studies/, prior reviews at research/reviews/, bibliography at research/bibliography/
   • The actual primary sources the target cites
   • .great-authors/project.md if cross-craft

3. Read the target. May be a single file, directory, or glob. Read in full — adjacent context, sibling files, the bibliography entries the target relies on.

4. Resolve the panel. If --personas given, use those. Otherwise default panel:

   • stephen-jay-gould-researcher — synthesis, essay-form integrity, just-so-story check
   • carl-sagan-researcher — skeptical inquiry, Baloney Detection Kit
   • edward-o-wilson-researcher — consilience check, biology-anchored synthesis

   Common alternative panels:

   • Civilizational-claim review: jared-diamond-researcher + edward-o-wilson-researcher — synthesis vs. consilience
   • Clinical / medical case review: oliver-sacks-researcher + atul-gawande-researcher — humanism vs. systems
   • Deep-research narrative review: rebecca-skloot-researcher + robert-caro-researcher — relationship-as-method vs. turn-every-page
   • Public-science communication review: carl-sagan-researcher + mary-roach-researcher — wonder vs. immersion
   • Sociobiology debate: edward-o-wilson-researcher + stephen-jay-gould-researcher — the canonical disagreement (group selection, adaptationism)
   • Cross-craft technical-rigor review: add great-engineers:don-knuth-engineer (cross-plugin)
   • Cross-craft political-philosophy review: add great-counsels:hannah-arendt-counsel (cross-plugin)

5. Dispatch the panel in parallel via the Agent tool. Each persona gets:

   • Full target content
   • Bible context
   • Specific instruction: Verdict (one sentence), Marked passages (3-8 quoted excerpts), Hand-off (if a different persona would serve better).

6. Consolidate. Output:

---
title: Research review of <path>
slug: <slug>
panel: [<persona-1>, <persona-2>, <persona-3>]
created: YYYY-MM-DD
target: <path>
target_lines: <N>
---

# Research review: <path>

> ⚠️ **Craft register only — not academic advice.** This review is a reasoning exercise from three persona lenses, not actual peer review.

## Per-persona verdicts

### <persona-1>
**Verdict:** <one sentence>

**Marked passages:**
- <quoted excerpt>
- ...

**Hand-off:** <if different persona would serve better; or omit>

### <persona-2>
[same]

### <persona-3>
[same]

## Where they agree

<1-3 points where the panel converges. Strongest signal.>

## Where they disagree

<1-2 points where the panel diverges. Often the most useful section. The Gould-vs-Wilson disagreement on adaptationism is canonical.>

## Highest-leverage change

<ONE recommendation. The single change that, if made, would make the study meaningfully better.>

## Suggested next step

<One of: implement the change, escalate to a domain expert, request a v2 from the study author, run /researchers-debate <topic> if the disagreement is structural, cross-dispatch great-engineers:don-knuth-engineer for technical rigor or great-counsels:hannah-arendt-counsel for political-philosophy framing.>

7. Save to research/reviews/<slug>.md.

8. Report:

   📝 Saved to research/reviews/<slug>.md (review of <path>, <word-count> words).
   ⚠️ Craft register only — not academic advice.
   
   Panel:           <persona-1>, <persona-2>, <persona-3>
   Convergence:     <one-line summary>
   Highest-leverage change: <one-line summary>
   
   Next:
   - Address the highest-leverage change
   - For real publication: actual peer review with domain experts
   

What the skill does NOT do

• Modify the target. It reviews; the author edits.
• Approve or reject for publication. It surfaces findings; the orchestrator decides.
• Replace journal peer review.
• Auto-pick the panel based on opaque heuristics. The default is documented; alternatives are documented; the user can override.

Notes

• Parallel dispatch means personas don't see each other's reviews when forming their own. Independent verdicts produce stronger signal at consolidation.
• For very large targets (>2,000 lines), the dispatch brief should ask each persona to focus on the most-critical sections only.
• The disclaimer block is not optional. Do not strip the "not academic advice" line from any review.