Guide · Agent skill
How to use the literature-search agent skill: setup and examples
Updated · written with AI from the project's own documentation and checked against our catalogue data
The literature-search skill is a methodology playbook for an AI agent. Once installed, it tells the agent how to plan a literature search, set screening criteria, screen papers in two passes and summarize findings with a PRISMA-style record. It is aimed at graduate students writing systematic or scoping reviews, researchers drafting background sections, and AI or ML teams that want documented search methods. This guide explains what the skill does, how to install it, how to trigger it and what to check in its output.
Literature Search Methodologyaiming-lab/AutoResearchClaw · MIT Download the skill Quick start
- 1Download and unzip. Download the literature-search ZIP from this page and unzip it. Keep the folder containing SKILL.md intact.
- 2Place the skill folder. Move it into ~/.claude/skills/ for personal use or into .claude/skills/ in your project. Alternatively, upload it under Settings → Capabilities → Skills in the Claude apps.
- 3Describe your review. Explain your research question, field and whether you are writing a systematic review, scoping review or background section.
- 4Ask for the PICO framing and concepts. Request a PICO question, two to four core concepts and synonym lists for each.
- 5Build and document search strings. Ask for Boolean queries for each database you plan to search, and save them exactly as written.
- 6Set criteria and plan screening. Have the agent draft inclusion and exclusion criteria and a two-pass screening plan before you review any papers.
- 7Verify and record. Check every citation in the real database and fill in your PRISMA-style counts from actual exports.
What does the literature-search skill make the agent do?
A skill is a folder containing a SKILL.md file with instructions that an AI agent reads when a task matches the skill's description. This one covers systematic literature review methodology: search strategy, screening and synthesis. Its instructions are a checklist, not a search engine.
The skill asks the agent to define the research question using the PICO framework (Population, Intervention, Comparison, Outcome), identify two to four core concepts, list synonyms for each, and combine them with Boolean operators (AND between concepts, OR within synonyms). It then points to databases by field: PubMed, Scopus or Web of Science for biomedical work; arXiv, Semantic Scholar, DBLP or ACL Anthology for computer science; and Google Scholar or OpenAlex for interdisciplinary work.
Install the skill from the ZIP download
Download the ZIP for the literature-search skill from this page and unzip it. You can install it in one of two places:
Personal install: unzip the folder into ~/.claude/skills/ so it is available across your projects.
Project install: unzip the folder into the .claude/skills/ directory inside a specific project so it applies only there.
In the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.
- Personal: ~/.claude/skills/
- Project: .claude/skills/ inside your project folder
- Claude apps: Settings → Capabilities → Skills (upload)
How do you trigger the skill with a natural request?
You do not need a special command. The skill's description says it applies when conducting literature reviews or writing background sections, so ask for that work in plain language. The file also lists trigger keywords such as literature, review, systematic, PRISMA, search, database, PubMed, arXiv and citation.
Naming the framework or the database in your request tends to help the agent pick the right methodology.
- Help me plan a systematic review on whether intermittent fasting affects insulin resistance, using PICO.
- Write a search strategy with Boolean strings for arXiv and Semantic Scholar on retrieval-augmented generation evaluation.
- Draft the background section for my paper on low-resource speech recognition and tell me how to document my search.
- Set up screening criteria and a PRISMA-style record for my scoping review.
What should you expect in the output?
Expect a plan, not a finished bibliography. Typical outputs are a PICO question, concept and synonym tables, Boolean search strings per database, a list of inclusion and exclusion criteria written before screening, a two-pass screening plan, a data extraction form and a synthesis outline with research gaps.
The skill also asks for quality checks by study type: Cochrane Risk of Bias for RCTs, Newcastle-Ottawa Scale for observational studies, and a check of reproducibility, dataset validity and statistical rigor for ML papers.
- PICO question and core concepts
- Synonyms and Boolean search strings per database
- Inclusion and exclusion criteria, set before screening
- Two-pass screening plan (title/abstract, then full text)
- PRISMA-style counts at each stage
What does the skill need, and what are the safety limits?
The source lists no required tools or packages. The skill runs no scripts and does not use the network. It contains only SKILL.md, so it does not connect to PubMed, arXiv or any other database.
This matters for accuracy. The agent relies on its own knowledge and any tools it already has. It may invent citations or details, so verify every paper it names in the database itself before you cite it. Its guidance is general and does not replace expert review or formal systematic review training, and it should not be used alone for medical or clinical decisions.
- Requires: nothing listed
- Runs scripts: no
- Uses network: no
- Always check citations against the original database record
Tips for getting reliable results
Give the agent your research question and domain up front, so it can choose the right databases and quality tools. Ask it to write out the exact search strings and inclusion criteria, because the skill says to document them for reproducibility. Decide your criteria before you look at results, since the skill specifies that criteria are set before screening begins.
Treat the PRISMA record as a template to fill with your real counts. Keep your own log of what was searched and when.
- State the domain (biomedical, computer science, interdisciplinary) at the start
- Ask for exact search strings you can rerun
- Set criteria before screening
- Fill the PRISMA counts from real database exports
What people use it for
- Planning a systematic review protocol for a graduate thesis, with PICO framing and documented search strings.
- Drafting the background section of a research paper and identifying gaps in the existing literature.
- Setting up a reproducible search across arXiv and Semantic Scholar for an ML research team.
- Designing a data extraction form and quality assessment plan for observational or RCT studies.
- Creating a PRISMA-style record template for a scoping review using the PRISMA-ScR extension.
Frequently asked questions
Is the literature-search skill free to use?
The source describes the skill as MIT licensed. It is a set of instructions in a single SKILL.md file, so no purchase is described on the source page. Check the repository's licence file for the exact terms.
What does the literature-search skill require to install?
The listing says it requires nothing beyond what is described. It runs no scripts and does not use the network, and the file contains only SKILL.md. You install it by unzipping it into ~/.claude/skills/ or .claude/skills/.
Will the skill search PubMed or arXiv for me?
No. The skill contains no search tools or scripts and does not connect to any database. It gives the agent a method for searching, and the agent may invent citations, so verify every result yourself.
Where do I install the literature-search skill?
Unzip it into ~/.claude/skills/ for personal use or into .claude/skills/ inside a project. In the Claude apps you can also upload it under Settings → Capabilities → Skills.
How is this different from just asking an AI for a literature review?
The skill gives a fixed, checklist-driven process: PICO framing, documented Boolean strings, criteria set before screening, two-pass screening and PRISMA-style counts. Without it, the agent may skip those steps unless you ask for each one.
Can I use this skill for medical decisions?
The guidance is general and does not replace expert review or formal systematic review training. Do not rely on it alone for clinical or medical decisions.
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 this skill if you are a graduate student, researcher or ML team that wants a documented, checklist-driven method for systematic reviews or background sections. Pick something else if you need an agent that automatically queries PubMed or arXiv and returns verified results, or if you need evidence-based clinical guidance; this skill provides method only and can produce invented citations.
Download the skill