Agentic Revision Automation

Automate your academic revisions.

A light-weight and powerful meta-prompting system for tracking comments and rewriting documents directly inside Claude Code.

When reviewers send back a .docx with dozens of comments, don't spend weeks matching context. R&R spawns a specialized team of subagents to extract, research, edit, and perfectly repack your document without formatting rot.

npm install -g revise-and-resubmit-cc
rnr-subsystem.exe

The Subagent Ecosystem

R&R doesn't rely on one massive AI. It uses a highly specialized assembly line of native Claude Code mini-AIs where each agent has exactly one precise job.

📦

The Extractor

Silently opens your .docx, finds every Reviewer comment and Edit Suggestion, extracts the exact paragraph attached to it, and saves them as isolated Markdown files.

⚙️

The Processors

Isolated & Interlaced. Dedicated agents spawned for each comment or interwoven cluster. They rewrite the text based strictly on your personal writing style, with empty clean contexts.

🔍

The Researcher

If a reviewer asks for a citation, the Processor spawns a Researcher agent which connects to your NotebookLM, finds the exact answer, and hands it back.

❓

The Clarifier

Encountered a vague comment? The Clarifier pops up in your terminal to explain the ambiguity and chat with you to figure out exactly what you want the Processor to do.

✨

The Assembler

Perfectly injects all the revised paragraphs back into a clean, formatted `.docx` file without messing up your citations, graphics, or margins.

Lean Orchestration

Zero Context Bleeding.

A major problem with using LLMs for full document revision is "Context Rot." If the AI reads a 50-page document and 100 comments, it will start forgetting instructions, ignoring style guides, and hallucinating facts.

  • Blind Orchestrator: The main coordinator never reads your manuscript. It just runs the workflow.
  • Native File Blocks: Subagents use native <files_to_read> blocks to ingest specific bits of context.
  • Fresh Context: Each edit happens in a completely fresh context window focused 100% on the single active comment.
  • XML Directed: Every agent is structured XML optimized specifically for Claude.
<classification type="needs_context">
  <reasoning>
    The reviewer asks us to cite the 
    2023 review by Johnson et al. 
    We need human input.
  </reasoning>
  <prompt_for_human>
    Please provide the full citation 
    and a brief summary for Johnson 
    et al. 2023.
  </prompt_for_human>
</classification>

The Workflow Commands

Simple `/rnr` commands built directly into your Claude Code terminal.

/rnr:init
Installs the parser and assembler scripts to a `.rnr/` folder in the current directory workspace.
/rnr:extract-comments <file>
Unpacks the `.docx` into XML and extracts Reviewer 1/2 comments as individual isolated files.
/rnr:synthesize-style
Crawls your manuscript to construct a custom style_skill.md from your vocabulary and tone.
/rnr:process-comments
The main engine. Iterates through each comment, spawning subagents and prompting for human context when needed.
/rnr:assemble <output>
Repacks all revised XML nodes flawlessly into a fresh, valid `.docx` ready for submission.