virtual-reading-group_skill

Overview

This skill orchestrates a multi-agent virtual academic reading group to analyze, discuss, and synthesize research across multiple papers. It runs configurable expert and junior researcher personas in four sequential phases to produce traceable notes, expert responses, and an integrated, theme-based synthesis with full citations. Outputs are written to a specified directory and scale from single-paper deep reads to 1–50 paper syntheses.

How this skill works

I spawn parallel expert-reader agents to read assigned papers through a user-provided research question, produce paper-level notes and session summaries, then run a junior researcher agent to generate probing questions and grand challenges. Experts respond to the junior’s questions in parallel, and a single synthesizer agent reads all prior outputs to produce a cohesive Integrated Discussion Summary organized by theme. Every factual claim and quote is tied to a paper citation and speaker attribution so outputs remain traceable.

When to use it

• You need coordinated, multi-perspective reading of a set of academic papers (1–50).
• You want expert-level discussion notes with traceable citations and quoted passages.
• You need a synthesized, theme-oriented summary combining disagreements, consensus, and open questions.
• You want configurable expert personas or to run iterative rounds of follow-up questions.
• You need outputs written to files for downstream workflows or archival.

Best practices

• Provide a clear research question to focus agents and improve signal-to-noise.
• Supply PDFs or text extracts and an output directory; pre-extract PDF text for problematic files.
• Limit 1–6 papers for single-expert runs; use multiple experts for larger sets (max ~6 papers per expert).
• Prefer opus for Phase 2 (junior) and Phase 4 (synthesis) for higher-quality reasoning; use sonnet for lower-cost reading phases.
• Enforce the citation and speaker-attribution format in prompts to maintain traceability.

Example use cases

• Synthesize 10 conference papers on a focused research question to produce publishable discussion notes and open problems.
• Run a weekly lab reading group: each paper assigned to an expert, produce session summaries and a consolidated weekly synthesis.
• Compare competing theoretical positions across a small corpus and extract testable hypotheses and methodological gaps.
• Archive structured discussion artifacts (notes, expert responses, integrated summary) for later review or grant proposal development.