A master's thesis in a fast-moving field and too many new papers. A weekly Codex automation searches arXiv, triages only new papers and files promising ones into Zotero, with fewer hallucinated citations.
Done withCodexSpotted· by Jonas Wiedermann-Möller· Oct 20, 2026
Result Sorts papers into keep, skim, drop or duplicate
“In my experience, this leads to fewer hallucinations during related work sections and less need to double-check citations.”
Spotted: found in the wild. Details are as reported by the source above and have not been independently verified.
What they did
A researcher working in a fast moving field needed to keep up with new literature without reading every paper that came out. They set up a weekly Codex automation that searches arXiv for papers matching chosen keywords and saves the results to a CSV so the same paper never gets triaged twice. Codex sorts each new paper into categories like keep, skim, drop, duplicate, outside scope or weak method, and promising ones get added to a Zotero inbox where they are automatically embedded in a vector database.
How it went
Codex can later search that Zotero collection against the full text, notes and metadata when drafting related work sections or checking a claim. The writer says this leads to fewer hallucinations and less need to double-check citations, while they still read the flagged papers themselves and keep responsibility for the research question.
Try it yourself with Codex
Set up a recurring automation that searches [arXiv or another paper database] for new papers matching your keywords, skips anything already seen using a saved CSV of past runs, and sorts each new paper into categories such as keep, skim, drop, duplicate, outside scope, or weak method. Add the promising papers to a reference manager collection so the agent can later search their full text and notes when you draft related work or need to check a citation.
A starting prompt written by Agents in the Wild. Fill in the [brackets], and check what the agent does before it sends, pays for or deletes anything.
Building their first AWS project by vibe coding, the writer designed the architecture and let Codex write the code. A flawed first design took a day, against an estimated three or four; the redesign also took a day and scaled.
A roughly 50-page tax return, run through three coding agents side by side. Codex was accurate first time in 12 minutes, Antigravity missed a value until prompted, and Claude Code stumbled overnight.
Four years after shelving a homemade RSS reader and selling off its $500 domain, the writer built it this year with Codex and Claude. They use it daily, along with 30 other strangers who found it public.
Nap-trapped by their baby, they built an MCP bridge so an agent could run their Ableton Live set directly. Codex composed an EDM track from scratch, then they revised vocals, drums and dynamics with follow up prompts.
A client wanted a Belmond hotel in Italy, and the advisor had only ever stayed at one Belmond, in Anguilla. Codex, pointed at the client folder, found the notes, checked Belmond's live site and ranked three options.
Six AI workers, two each on Claude, Codex and Gemini, were set loose to build a SQLite-like engine in Rust. It passed 64 queries of a 1,000+ query suite, with slop like redundant buffer copies and no concurrency.