Paper Context Resolver

When you're reproducing an AI paper from a GitHub repo and hit a specific gap the README can't fill, this resolves narrow technical details from the original paper sources. It handles dataset splits, preprocessing steps, evaluation protocols, and checkpoint mappings that repos often assume you know. The skill explicitly flags conflicts between README instructions and paper methodology, so you're not left guessing which approach to follow. It's designed as a targeted helper, not a paper summarizer, and outputs references to paper-assisted-reproduction.md for documentation. Best for bridging concrete reproduction gaps rather than general paper understanding.