Claude Skill MartBrowse skillsQuick linesLearn by videoTerminal guideWhat is a Skill?
 
← Back to list

Content Refinement Agent (Step 5)

An iterative LaTeX paper refinement loop that simulates peer review and only keeps revisions that provably raise the score.

Content & WritingAdvanced★ 668⑂ 93AI score 8/10Last updated: Sep 21, 2026

What it does

  • Runs a simulated peer review on workspace/drafts/paper.tex, scoring six axes (scientific depth, technical execution, logical flow, writing clarity, evidence presentation, academic style).
  • Applies targeted revisions from the reviewer feedback and snapshots every iteration (iter1/, iter2/, ...) with compiled PDFs.
  • Decides ACCEPT / REVERT / plateau halt / target-met halt (≥80) deterministically via score_delta.py exit codes, with real rollback when the score regresses.
  • Pre-flight gates: 7 AI failure modes, claim-evidence provenance for every number, reverse outline of topic sentences, and a research-brief read.
  • A designated Devil's Advocate reviewer attacks core claims; an unresolved CRITICAL finding blocks acceptance even if scores improve.
  • Iron rules against reward hacking: no new experiments, no new numbers, no defensive limitation-listing.

Who it's for

  • Researchers and grad students polishing LaTeX submissions for conferences.
  • Users running the full PaperOrchestra pipeline (Steps 1–5).
  • Anyone who wants automated editing with a genuine undo path.

Examples

  1. "Run peer review on this paper" → review.json, per-axis scores, and a decision band (Accept/Minor/Major/Reject).
  2. "Iterate on the paper three times" → iter1–iter3 snapshots, an append-only worklog.json, and the best snapshot promoted to workspace/final/.
  3. "Iteration 2 scored lower — what happened?" → automatic REVERT to iter1 plus a reported halt reason.

· · · Install guide · · ·

Try it now, no install

Paste this into Claude to use the skill without installing anything.

Read the instructions in this file and follow them to help me:
https://raw.githubusercontent.com/Ar9av/PaperOrchestra/HEAD/skills/content-refinement-agent/SKILL.md

What I want: (describe your task here)

If Claude can't open the link, open it yourself and paste the contents instead.

↓ If it works for you, download the ZIP below and install it. Then it runs on its own — no pasting each time.

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the skills/content-refinement-agent folder from the GitHub repo Ar9av/PaperOrchestra into my ~/.claude/skills/content-refinement-agent/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/Ar9av/PaperOrchestra.git && mkdir -p ~/.claude/skills && cp -r PaperOrchestra/skills/* ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

  1. Open a terminal in your working directory.
  2. Clone the repo: git clone https://github.com/Ar9av/PaperOrchestra.git
  3. Create the skills directory: mkdir -p ~/.claude/skills
  4. Because this skill references skills/shared/ and skills/paper-orchestra/scripts/, copy the whole bundle: cp -r PaperOrchestra/skills/* ~/.claude/skills/
  5. Ensure LaTeX is available for compilation: run latexmk -v; install TeX Live or MacTeX if missing.
  6. Verify Python 3: python3 --version
  7. Prepare inputs in your project: workspace/drafts/paper.tex, workspace/inputs/conference_guidelines.md, workspace/inputs/experimental_log.md.
  8. Launch Claude Code and say "refine the draft" or "run peer review on this paper" to trigger the skill.
View source on GitHub ↗License: NOASSERTION