Literature Review Assistant

Problem

A literature review often stalls after a broad search returns many papers but no clear way to decide which ones support the claim. This skill targets engineers and researchers who need a reviewable workflow rather than a list of titles. It frames the task as scoping the question, designing the search, screening sources, extracting evidence, synthesizing themes, and verifying citations.

How It Works

• Question scoping: confirm the object, intervention or phenomenon, outcomes, context, and time boundary, then state what the current pass will not answer.
• Search design: build Chinese and English synonym sets, abbreviations, spelling variants, and Boolean query strings; record the database, date, full query, and hit count.
• Screening and extraction: apply title, abstract, and full-text inclusion rules with explicit exclusion reasons; organize metadata such as title, year, design, sample, results, limitations, and DOI.
• Thematic synthesis: group findings by mechanism or research question, showing consensus, counterexamples, evidence strength, and applicable conditions instead of summarizing paper by paper.
• Citation verification: check authors, year, volume/issue/pages, DOI, and links, and mark any item that remains unverified.

Boundaries

It does not bypass paywalls, logins, captchas, or institutional access controls. It does not describe abstract-level reading as full-text review, and it does not treat preprints as completed peer-reviewed evidence. When databases are unavailable, it can deliver the search plan, templates, and a to-do list, clearly marked as not yet executed.