What problem does it solve?
It streamlines the end-to-end process of turning an ML research idea into a submission-ready paper by coordinating iterative experiment design, evidence-grounded writing, citation verification, and revision based on feedback.
Core Features & Use Cases
- Venue-targeted paper production: Covers the full lifecycle for NeurIPS, ICML, ICLR, ACL, AAAI, and COLM with an explicit iterative loop across design, experiments, drafting, review, and submission.
- Experiment-to-claim discipline: Maps each paper claim to specific experiments and enforces that results support narrative structure.
- Citation integrity: Prevents hallucinated citations by requiring programmatic lookup and marking unverifiable items as CITATION NEEDED.
- Experiment operations at scale: Recommends robust monitoring, crash-safe result saving, journal-style exploration tracking, and statistical analysis patterns.
- Submission readiness support: Includes conference-relevant expectations (e.g., checklists and reporting standards) and guidance for formatting/figures/tables.
Quick Start
Use the research-paper-writing skill to create a NeurIPS submission draft by first defining a single-sentence contribution, then running claim-mapped experiments, then writing an evidence-grounded first paper draft for iterative self-review.