What problem does it solve? Academic posters built by hand or by unguarded AI generation tend to end up with inconsistent colors, missing real paper figures, broken layouts, and endless cosmetic patch loops. This Skill replaces eyeballing with hard measurement gates so a poster only ships when alignment, style, figure provenance, and print output all pass. ## Core Features & Use Cases - Measurement-gated pipeline: Runs preflight, style, asset, measure, and polish gates in a canonical order via run_gates.py, producing a GATE_REPORT.json ledger before any aesthetic review happens. - Real figure provenance: Extracts figures from the paper PDF at 300-450 DPI, tracks them in FIGURE_MANIFEST.json with sha256 hashes, and hard-fails posters with fewer than two genuine paper figures. - Two-hue design token system: Enforces a single accent plus gold palette through token packs (generic, ICLR, ICML, NeurIPS, CVPR, ACL) and a 12-rule style gate that bans inline styles, gradients, and stray hex colors. - Use Case: Given an ICLR paper PDF, produce a 185x90cm landscape poster.html, iterate until all gates pass and the visual review scores 9/10, then export a verified single-page print PDF under 20MB. ## Quick Start Ask the AI to make a conference poster from your paper PDF for a specific venue, for example: create an ICLR 2026 poster from paper.pdf on a 185x90cm landscape canvas.
