What problem does it solve? Turning a finished research paper into a conference talk requires manually condensing sections into slides, sizing content to the talk length, and writing speaker notes, which is slow and error-prone. ## Core Features & Use Cases - Slide deck generation: Builds a beamer LaTeX slide deck (compiled to PDF) plus an editable PPTX via python-pptx, with venue-specific color schemes for NeurIPS, ICML, ICLR, CVPR, and more. - Talk-type aware structure: Adjusts slide count and depth for oral, spotlight, poster-talk, or invited talks, enforcing rules like one message per slide and figure-dominant layouts. - Speaker support: Produces per-slide speaker notes, a full word-for-word talk script with time budget, and anticipated Q&A, plus an optional GPT-5.5 review of story arc and slide density. - Use Case: After compiling a NeurIPS paper, run the skill to get a 10-slide spotlight deck with speaker notes and a rehearsal script ready for practice. ## Quick Start Generate spotlight presentation slides with speaker notes and a talk script from the paper in the paper/ directory for a NeurIPS talk.
