Skip to main content
subagent CX Codex

paper-reviewer — Claude Code subagent: adversarial paper review

paper-reviewer — a reviewer subagent for a co-mathematician setup. It checks reports, citations, proofs and code, and a workstream stays open until it writes an approval file.

no license ★ 240 updated 2026-09-18

Install into Claude Code · Manually: put the files in place All methods and platforms ↓
.claude/agents/paper-reviewer.md
Open on GitHub

What paper-reviewer does

This agent is built to disagree. It reads report.md, log.md and the matching part of paper.tex, then records APPROVE, REQUEST_CHANGES or REJECT in .co-math/approvals/<workstream-id>-<round>.md. No Verdict: APPROVE line, no move to complete.

What it inspects follows the workstream type. Citations get matched against references/<id>/note.md, and at least one per round is fetched with WebFetch to see whether the paper sits where the note claims. Proof work is walked step by step; anything waved through as obvious is blocking, and under strict_mode: true in co-math-config.json phrases like clearly or omitted are treated the same unless a justification or \unproven{} follows. Code work gets pytest -q tests/ run from the workstream root, plus a look at whether golden values were fitted backwards to whatever the program printed.

Lean results go through plumbing checks rather than a second proof attempt, since the compiler already settled the maths. The agent rebuilds with lake build, greps Main.lean for a live sorry, and compares the formal statement against the claim in paper.tex. A formal statement weaker than the paper's is blocking. Tools it can reach for: Read, Write, Edit, Bash, Glob, Grep, WebFetch. Citation spot-checks need network access, and a first mathlib build takes a long while before it either passes or fails.

What it helps with

  • keep a workstream open until the reviewer signs off
  • catch a citation pointing at a paper that does not exist
  • expose a proof step asserted as obvious with nothing behind it
  • trace numbers in the paper back to a test run
  • confirm a Lean theorem states what the paper claims

Example prompts for the agent

  • Review workstream lit-3, its status flipped to review, then write the approval file.
  • Re-check section 4 of paper.tex against the prover's report and list the blocking findings.
  • The Lean build for thm-main is stuck in review, run the check again and give a verdict.

How to install paper-reviewer

Manually: put the files in place

.claude/agents/paper-reviewer.md

For all projects use ~/.claude/agents/. Claude delegates by the description field or an @-mention

Step by step, with commands for each platform: Claude Code subagents: how to create one, call it and keep context clean

Questions about paper-reviewer

How do I install paper-reviewer in Claude Code?

Put the agent file into your project at .claude/agents/paper-reviewer.md.

What does paper-reviewer do?

It reviews a workstream's report: whether citations resolve, whether proofs carry leftover \unproven blocks, whether code output matches the numbers claimed. Each round ends with a verdict file in .co-math/approvals/.

Why is a Lean proof reviewed differently?

The compiler has already checked the mathematics, so the agent verifies the machine side instead: lake build exits 0, no live sorry remains in Main.lean, and the formal statement matches the theorem in paper.tex.

Similar subagents

All subagents for Science and research →

From the same set98

The whole set →
project-coordinator project-coordinator is a Claude Code subagent that coordinates mathematics research projects. It discusses goals, formalizes research intent, and delegates work to specialized sub-agents. CX Codex ★ 240repo stars good-story good-story — a skill that turns a manuscript, grant or abstract into one coherent research story and checks whether the evidence carries it. CX Codex +1 ★ 240repo stars 02a-empirical-reality-check 02a-empirical-reality-check audits real-world market assumptions prior to economic modeling, validating institutional claims against live web sources. GP ChatGPT ★ 240repo stars 10-manuscript-skeleton The 10-manuscript-skeleton prompt turns earlier economics research outputs into a manuscript outline, covering the abstract, model, results, proofs, discussion, and revisions. GP ChatGPT ★ 240repo stars fred-economic-data fred-economic-data pulls economic time series from the FRED API into Claude Code, including GDP, inflation rates, and payroll numbers. CX Codex +1 ★ 240repo stars experiment-claim-audit experiment-claim-audit is a Claude Code skill that checks paper numbers, comparisons, and scope claims against raw results. Run it before submission to verify reported values. CX Codex +1 ★ 240repo stars nature-story nature-story — a skill that transforms raw research ideas into rigorous scientific narratives with competing hypotheses, literature benchmarks, and figure-driven paper plans. CX Codex +1 ★ 240repo stars referee-report Draft or audit referee reports for economics and finance papers. Use it to assess a paper’s contribution and distinguish publication-critical issues from optional suggestions. CX Codex +1 ★ 240repo stars

Project sponsors

Companies that keep this analytics open