literature-review

Orchestrate literature reviews by processing PDFs through isolated subagents.

1|Updated Jan 1, 2026
One-click install
npx skills add https://github.com/sxg/science --skill literature-review-sxg
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: literature-review
Source: https://github.com/sxg/science/tree/main/plugins/writer/skills/literature-review
Command: npx skills add https://github.com/sxg/science --skill literature-review-sxg

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill helps researchers coordinate and manage literature reviews by orchestrating per-paper analysis through isolated subagents, preventing context overflow and enabling scalable synthesis.

Core Features & Use Cases

  • Isolated subagent processing: Each paper is processed by a dedicated subagent, generating concise notes and metadata.
  • End-to-end workflow: From scope context to bibliography, synthesis, and Introduction draft.
  • Scalable collaboration: Supports large literature sets with per-paper processing and centralized synthesis outputs.
  • Use Case: A project with 40 PDFs can be processed sequentially to produce a coherent Introduction without overwhelming a single context.

Quick Start

Provide the scope.md and papers directory and run the literature-review workflow to start processing papers one by one using the orchestrator.

Frequently Asked Questions about literature-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I manage context overflow when analyzing a large number of research PDFs?

To manage context overflow when analyzing research PDFs, this skill processes each paper through an isolated subagent, preventing the orchestrator from reading PDFs directly and ensuring scalable synthesis for large literature sets.

What is the best way to automate literature review synthesis and bibliography generation?

The best way to automate literature review synthesis is using an orchestrator that propagates scope context to dedicated subagents, enabling per-paper notes, centralized bibliography generation, and a final Introduction draft.

Can I process 40 PDFs sequentially to draft a coherent research introduction?

Yes, you can process 40 PDFs sequentially to draft a coherent introduction. The workflow scales by assigning dedicated subagents to generate per-paper notes, which are then centralized for final synthesis without overwhelming a single context.

How do I start a literature review workflow from a directory of papers?

To start a literature review workflow, provide a scope.md file and a papers directory, then run the orchestrator to begin processing papers one by one through isolated subagents to generate notes and synthesis.

Does the orchestrator need to directly read PDFs to generate per-paper notes?

No, the orchestrator does not directly read PDFs to generate per-paper notes. It strictly delegates PDF processing to isolated subagents, ensuring structured context propagation from scope.md while preventing context overflow.