xiongqi123123/awesome-rebuttal

Awesome Rebuttal is an add-on skill package for AI coding assistants (specifically Codex, Claude Code, and Cursor) that helps academic researchers write the author response when a submitted paper receives reviewer feedback from a conference or journal. After a paper is reviewed, authors typically get a short window to respond to each reviewer's concerns, and that process (called a rebuttal) requires careful argumentation, evidence, and formatting. This package gives an AI assistant a structured way to help with that task. Once installed, the package lets you ask your AI assistant to set up a rebuttal workspace in your paper's project folder. It will organize reviewer comments into atomic concerns, help you decide which concerns need new experiments versus which can be addressed with existing evidence, and draft responses in the appropriate format for the venue, whether that is a one-page PDF, per-reviewer replies on a platform like OpenReview, or a global comment section. The package includes safety rules to prevent the assistant from inventing experiment numbers, fabricating citations, or making claims that are not backed by evidence you have already provided. It will also not use a hostile tone toward reviewers or leak author identity in a double-blind venue. These constraints are baked into the skill definition so the assistant enforces them automatically during rebuttal drafting. All working data (drafts, notes, memory of reviewer positions, snapshots of drafts at different stages) stays in a hidden folder called .awesome-rebuttal/ inside your workspace. Nothing is sent to external services by this package itself. Installation involves cloning the repository and copying the files into the skills directory that your AI assistant reads, with slightly different paths for each supported assistant. The README includes step-by-step commands for each option. The package also includes a fallback LaTeX template for conferences that require a one-page rebuttal document.