Revise

SkillAI & models

R&R cycle, classify referee comments and route to appropriate agents. Replaces /respond-to-referee.

Instructions available. Your AI can read the instructions. Execution depends on the setup they require.

Add ahel to your AI once: Claude, ChatGPT, Cursor, Claude Code or Codex. Then ask it to use this.

Then ask your AI: use the Revise skill

What this skill tells your AI

The instructions your AI receives, as published by brycewang-stanford/auto-empirical-research-skills in skills/16-hsantanna88-clo-author/dot-claude/skills/revise/SKILL.md and read by ahel’s review.

Structure point-by-point referee responses with classification, agent routing per revision protocol, and diplomatic drafting.

Input: $ARGUMENTS — path to referee report file(s), optionally followed by paper path.


Workflow

Step 1: Parse Inputs

  1. Read referee report(s) from $ARGUMENTS
  2. Read the paper (paper/main.tex or specified path)
  3. Read revision protocol from rules
  4. Read existing scripts to know what analyses already exist

Step 2: Classify Every Comment

ClassRoutingAction
NEW ANALYSIS→ Coder agentFlag for user, create analysis task
CLARIFICATION→ Writer agentDraft rewritten section
REWRITE→ Writer agentDraft structural revision
DISAGREE→ User (mandatory)Draft diplomatic pushback, flag for review
MINOR→ Writer agentDraft fix directly

Step 3: Build Tracking Document

Save to quality_reports/referee_response_tracker.md with:

  • Summary counts per referee
  • Action items by priority (HIGH: new analysis, MEDIUM: clarification, FLAGGED: disagreements, LOW: minor)

Step 4: Dispatch Agents

  • CLARIFICATION/REWRITE → dispatch Writer with specific instructions
  • NEW ANALYSIS → flag for user approval before dispatching Coder
  • DISAGREE → draft diplomatic response, flag prominently for user

Step 5: Draft Response Letter

Generate LaTeX response letter with:

  • Summary of major changes
  • Point-by-point responses with exact referee quotes
  • Color-coded responses
  • Page/section references for each change

Step 6: Diplomatic Disagreement Protocol

When DISAGREE: open with acknowledgment, provide evidence, offer partial concession, NEVER say "the referee is wrong." FLAG for user review.

Step 7: Save Outputs

  1. Tracker: quality_reports/referee_response_tracker.md
  2. Response letter: quality_reports/referee_response_[journal]_[date].tex
  3. Revised sections: paper/sections/ (for CLARIFICATION/REWRITE items)

Principles

  • The response letter is the user's voice. Match their tone.
  • Never fabricate results. Mark NEW ANALYSIS items as TBD.
  • Flag all DISAGREE items. These need human judgment.
  • Track everything. Every comment appears in both tracker and response letter.

Signals

GitHub stars
4k
Forks
531
Last commit
Sep 2026
Advanced
Item type
skill
Key
revise-brycewang-stanford
Source
github.com/brycewang-stanford/auto-empirical-research-skills