Med Literature Tracker

Problem

Medical literature changes quickly, and scanning only titles or abstracts can miss relevant studies. Med Literature Tracker focuses on keyword-based literature tracking: it confirms journal scope, time window, and update cadence, then retrieves, ranks, extracts abstracts, and summarizes selected papers into a readable briefing. This is useful for turning scattered candidates from PubMed, Web of Science, and domestic core journals into a structured output.

Workflow and boundaries

The flow starts with parameters. English keywords are more stable; Chinese keywords are translated before retrieval. By default it targets Tier 1 and Tier 2 journals, with domestic core journals optional. PubMed is prioritized because it is public, while Web of Science is used as a supplement. Candidate papers are ranked by journal tier, keyword match, recency, and study design, and the top 10 to 15 are kept. For each selected paper, the skill extracts Title, Authors, Journal, Abstract, DOI, and PMID, translates the abstract while preserving statistical values such as OR, HR, 95%CI, and p, displays the briefing in chat, and writes a Markdown file to the output directory. If periodic updates are requested, it can configure daily, weekly, or monthly cron tasks.
Boundary: the skill is abstract-first and does not guarantee full-text access. When WoS is restricted, it falls back to PubMed results and notes the source limitation.