What problem does it solve? Peer-reviewing academic papers is time-consuming and error-prone: fabricated references, inconsistent numbers, and AI-generated writing are easy to miss, and reviewers must also respect confidentiality rules about sending manuscripts to external AI services. This Skill structures the entire review process for speech and language processing venues (Interspeech, ICASSP, IEEE/ACM Trans., Speech Communication) so nothing is overlooked. ## Core Features & Use Cases - Pre-review compliance hearing: Confirms whether the work is a committee review or self-review, checks the venue's AI-use policy, and enforces confidentiality rules (no manuscript content sent to external services without permission). - Three parallel review tracks: Technical content review (novelty, baselines, numeric consistency checks), reference verification (hallucination detection via DOI resolution and multi-database lookup), and language/AI-writing-trace analysis based on corpus-measured style data. - Structured deliverables: Produces per-track reports, a final review draft with proposed scores and submission-ready comments, and a self-contained HTML report for the reviewer. - Use Case: A reviewer assigned three ICASSP papers runs all three check tracks per paper in parallel, receives verified reference tables flagging two fabricated citations with DOI evidence, and gets a final review draft whose comments are scrubbed of AI-sounding phrasing before submission. ## Quick Start Ask the AI to review the attached manuscript PDF using the paper_review skill, stating whether it is a committee review or a self-review of your own draft.
