neel-nandaify-paper

You audit an ML research draft by applying a verifiable checklist that links each tick to specific prose in your current LaTeX document. This grounds writing best practices into a review process for catching weak claims and reproducibility gaps.

A paper audit checklist is a structured Markdown list based on Neel Nanda's ML writing advice. It helps researchers catch missing evidence, unclear exposition, and reproducibility gaps by anchoring every completed check to exact sentences in the current draft.

You set up a pre-submission review by copying a Markdown checklist template into your paper directory. You then work through it item by item, anchoring each completed check to the exact sentence or label in your current .tex draft.

Yes, you can use the checklist approach for revising existing research papers. The process supports revision passes by tying every completed item to specific prose in the draft so reviews stay grounded in the current text.

Yes, you need a root SKILL.md with YAML frontmatter and a Markdown checklist. The checklist must link each tick to specific prose in your current .tex draft to ensure the review remains verifiable.

The best way to track evidence and claims in LaTeX drafts is using a living paper checklist that anchors each completed check to specific text. This prevents weak claims and missing evidence from being remembered loosely instead of verified.