Screen research papers against criteria
Codex reads a folder of PDFs to find articles that disagree with an argument, screens about fifty papers for a systematic review, and cleans messy CSV or Excel files.
Done withCodex
- What they did
- User installs Codex app, opens a project folder containing PDFs or CSVs, and creates or lets Codex generate an AGENTS.md file with role, standards, and task instructions. Codex is given access to read all files in the folder, then prompted to screen papers against inclusion/exclusion criteria, identify articles disagreeing with a pasted argument, or clean a messy CSV/Excel file; it extracts information, builds tables or new files, and requests permission for actions like renaming or saving.
- How it went
- Codex outputs results as tables or new markdown/Word/Excel files in the folder. User must still verify screening and results before relying on them, as Codex handles labor-intensive extraction but not final judgment.
- Worth knowing
- Set Codex to ask permission before editing files; results require manual verification.
Try it yourself with Codex
Read all the PDFs in [folder] and screen them against these criteria: [inclusion and exclusion criteria, or the argument to test]. Produce a table with each paper, include or exclude, and a one-line reason with a quote. Finish when every paper is covered, and flag unclear ones for me rather than guessing.
Source: X · Undated
Five of these in your inbox every morning
The best things people got an AI agent to do, each with the prompt to try it.
More like this
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.
Recommend a hotel from client notes using Codex
A travel advisor had Codex use client notes to recommend a Belmond hotel in Italy, checking live booking engines and generating a detailed Word doc.
Build a racing game from a single prompt
Codex made a racing game with multiple racers, eight maps and items from one prompt, acting as designer, developer and QA tester by playing it.
Find connections across 100 non-fiction books
Claude Code browses a personal library with CLI tools and surfaces trails of related excerpts across books.
- Custom agent
Search literature and write notes into Zotero
A researcher uses Claude to search Google Scholar, follow citations, query a Zotero library in natural language, and write summary notes into Zotero or Obsidian.
Grade Cursor on a real data science project
A data scientist tested Cursor on a real project: weak on pandas, data cleansing and scraping, strong at prototyping.