Write a 45-page medical imaging literature review
The author had Claude Code use arXiv/PubMed searches and supplied papers to write a ~13,800-word review with 113 citations, noting that good source papers mattered most. ·
Done withClaude Code
- What they did
- In plan mode, Claude Code read one or two top-journal reviews to produce a style guide (CLAUDE.md). It then pulled the author's paywalled papers from Zotero through a Zotero MCP and searched arXiv and PubMed through MCPs. It drafted a review plan (IMPLEMENT.md), which the author checked by hand. Claude Code then wrote the draft without intervention.
- How it went
- The output was 45 pages, 13,813 words and 113 citations, all from arXiv, PubMed or the supplied papers, in Nature style. A plain one-line prompt gave a weak review. The author still has to review the draft and add figures.
- Worth knowing
- Paywalled papers must be downloaded by hand into Zotero. Don't point Claude Code at a local folder of PDFs, because a few papers can overflow the context.
Try it yourself with Claude Code
Write a literature review of about [word count] words on [topic] using the papers in [folder] plus searches of arXiv and PubMed for gaps. Cite every claim with a numbered reference, and give me a reference list with links. When finished, list any claims you couldn't source well so I can check them.
Want help getting this working? Join the Vibe Coding Academy ↗ Run by Alex Finn, who also runs this site.
1 more account of this
Source: agent.csdn.net · Nov 11, 2026 · Submitted by from search gap: 文献
Related use cases
Turn tickets into merged pull requests
Pulls tasks from GitHub Issues or Linear, then runs Claude Code or Codex in isolated Kubernetes pods to carry each one through to a PR.
Diagnose a killed process on a server
Codex CLI over SSH found the OOM killer in kernel logs on a 2 GB droplet with no swap and listed fixes by risk without changing anything.
Manage several websites across WordPress, Astro and Eleventy
The poster uses OpenClaw, alongside Claude Code and Codex, to manage and work across about five websites.
Run persistent financial research with an agent
LangAlpha is a Claude Code-style agent for investing research that keeps persistent workspaces and memory so analysis such as models and comps can be updated across sessions.
Custom agent
Generate deep research reports into Notion
An n8n workflow turns a topic into up to three Tavily searches, compiles a Markdown report with an agent, and pushes it to a Notion page.
Run a weekly competitive-monitoring briefing for product management
A product manager had Hermes produce a weekly competitor briefing; its skill rewrote itself four times and by week six surfaced patterns he had missed manually.
Get 5 like this every morning
The best things people got an AI agent to do, each with the prompt to try it.