Skip to content
Get packs

Guide · Agent skill

How to use the Research Literature Review agent skill (research-lit)

Updated · written with AI from the project's own documentation and checked against our catalogue data

The research-lit skill is a set of instructions that teaches an AI coding agent how to search for academic papers, read the first pages of local PDFs, and summarize what it finds. This guide explains what the skill makes the agent do, how to install it, how to trigger it with everyday requests, and what to check before using it on private material. By the end you will be able to run a literature search and know which sources were used.

Research Literature Reviewwanshuiyin/Auto-claude-code-research-in-sleep · MIT

Quick start

  1. 1Download the ZIP. Get the research-lit ZIP from this page. It contains SKILL.md.
  2. 2Unzip into a skills folder. Place the folder in ~/.claude/skills/ for personal use or .claude/skills/ inside a project.
  3. 3Add your PDFs (optional). Put PDFs in a papers/ or literature/ folder in your project.
  4. 4Connect optional tools (optional). If you use Zotero or Obsidian, connect their MCP servers so the agent can search them.
  5. 5Ask for papers. Write a request such as find papers on a topic or literature review on a topic.
  6. 6Choose sources if needed. Add a sources directive to limit or extend the search.
  7. 7Review and verify. Read the summaries and check each citation against the original paper.

What the research-lit skill makes the agent do

Agent skills are folders of instructions that an agent such as Claude loads when a task matches the skill's description. The research-lit skill is described as a way to search and analyze research papers, find related work, and summarize key ideas.

When you ask for papers on a topic, the skill tells the agent to search several sources in a set order: your local PDF folders, Zotero and Obsidian if they are connected, and then the web, arXiv and optional academic APIs. It reads the first few pages of local PDFs and collects titles, abstracts and authors, then summarizes the key ideas for you.

  • Searches local PDFs in papers/ or literature/ folders, reading the first 3 pages of up to 20 files.
  • Searches Zotero and Obsidian when their MCP servers are configured.
  • Searches the web and arXiv, collecting metadata.
  • Summarizes key ideas from what it found.

Install the research-lit skill

Download the ZIP for the skill from this page and unzip it. The skill is a folder containing a file named SKILL.md.

For a personal install that works in every project, unzip the folder into ~/.claude/skills/. For a single project, unzip it into the project's .claude/skills/ folder.

In the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.

  • Personal: ~/.claude/skills/
  • Project only: .claude/skills/ inside your project folder
  • Claude apps: Settings → Capabilities → Skills

Trigger the skill with natural requests

You do not need a special command. The skill's description lists phrases that should start it, so ask in plain language. The skill also accepts a topic or URL as an argument, shown as [paper-topic-or-url] in its header.

Phrases the description names include "find papers", "related work", "literature review", "what does this paper say", and requests to understand academic papers.

  • "Find papers on retrieval-augmented generation for code search."
  • "I need a related work section on graph neural networks for molecules."
  • "Do a literature review on diffusion models for audio."
  • "What does this paper say about its training data?"

Choose which sources to search

By default the skill searches all of its default sources. You can narrow the search by adding a sources directive to your request. The skill's file lists the valid source names: zotero, obsidian, local, web, semantic-scholar, deepxiv, exa, gemini, openalex and all.

Semantic Scholar, DeepXiv, Exa, Gemini and OpenAlex only run when you name them explicitly. Other sources are skipped silently if they are not configured.

  • Only Zotero and local PDFs: sources: zotero, local
  • Only the web: sources: web
  • Web plus Semantic Scholar for published venue papers: sources: web, semantic-scholar
/research-lit "diffusion models" — sources: zotero, local

Set PDF library location and arXiv downloads

The skill checks papers/ first, then literature/ in your current project directory. You can point it at another folder by adding a Paper Library section to CLAUDE.md with the custom path.

By default the skill only fetches arXiv metadata (title, abstract, authors) and downloads no files. Turning on arXiv downloads is optional and must be requested.

  • Custom library path: add a ## Paper Library section to CLAUDE.md.
  • Download top arXiv PDFs: arxiv download: true
  • Set the download limit: max download: 10
/research-lit "topic" — arxiv download: true, max download: 10

Safety notes before you run it

The skill runs scripts and uses the network. Some optional sources send your queries to outside services. Those queries leave your machine, so do not use these sources if you do not want that.

The skill is not a citation checker. Its source file does not describe checks that confirm every reference is real or accurately quoted, so verify citations yourself before publishing.

  • Keep large private collections out of optional web or AI-discovery sources if you do not want PDF content sent to those services.
  • Optional API keys are only used when you explicitly request those sources.
  • Check every reference against the original paper before you cite it.

What people use it for

  • Drafting the related work section of a paper by collecting titles, abstracts and authors on a topic.
  • Getting a quick explanation of what a specific PDF in your folder says.
  • Searching a Zotero collection or Obsidian notes for papers you already saved.
  • Collecting arXiv metadata to scan a field before choosing papers to read.

Frequently asked questions

How do I install the research-lit skill?

Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. In the Claude apps you can also upload it under Settings → Capabilities → Skills.

Is the research-lit skill free?

The skill is released under the MIT licence. Optional sources such as Semantic Scholar, Exa or Gemini may need their own API keys, which this page does not cover pricing for.

What do I need to use research-lit?

Nothing beyond the skill itself is strictly required, since the web and arXiv metadata sources work without extra setup. Zotero, Obsidian, PDFs in papers/ or literature/, and API keys are optional.

Does research-lit download papers automatically?

No. By default it only fetches arXiv metadata and downloads no files. You must turn on arXiv downloads with arxiv download: true.

Can research-lit verify that citations are real?

The skill's source does not describe checks that confirm every reference is real or accurately quoted. Treat its output as a starting point and verify citations against the original papers.

How is this different from other literature search tools?

This skill runs inside an agent and combines your local PDFs, Zotero, Obsidian and web sources in a set order. Dedicated search websites may offer richer browsing, so pick the tool that fits your workflow.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Use research-lit if you are starting a literature review, need to understand a specific paper, or already keep a Zotero or Obsidian library you want the agent to search. Pick something else if you need verified citations or you cannot send any query or PDF content to outside services.

Download the skill