Scholarly Source Fetcher

A fetcher skill that distills peer-reviewed scientific sources into a single structured dataset. It queries the Google Scholar, PubMed, Semantic Scholar, and OpenAlex APIs in parallel to pull papers relevant to your topic or research question; for each record it returns the DOI, title, authors, journal, publication year, citation count, study type (RCT / meta-analysis / systematic review / cohort), sample size, open-access status, and a relevance score between 0 and 1.

The skill prioritizes meta-analyses and systematic reviews in its ranking, detects duplicate records (for example, the preprint and published versions of the same study) via DOI and title similarity, and flags them in the duplicate_of field so your literature review stays clean.

When to use it: When kicking off a literature review, building a source pool before tying a hypothesis to a chain of evidence, or quickly gathering high-citation studies that support a claim. It also runs as a worker skill for literature-review-scout.

Output: A JSON array of rows with consistent keys (one row = one paper). It can be fed directly into a table, a Notion database, or a downstream analysis step. You can sort by relevance score and citation count to surface the strongest evidence at the top.