What problem does it solve? Grading dozens of student Quarto or R Markdown submissions question by question is slow and repetitive. This Skill performs a headless first-pass grading across all repositories in a markermd project, pre-toggling rubric items with evidence notes so the human grader only reviews and corrects instead of starting from scratch. ## Core Features & Use Cases - Headless rubric application: Reads each repository's per-question sections, decides which rubric items apply (matching descriptions verbatim), and records reasoning as private notes. - Schema-validated marks YAML: Writes a marks file in format_version 1.0, validates it against the bundled JSON Schema, and imports it with marks_import() without clobbering existing grading. - Human-in-the-loop review: Shows the marks and per-question score arithmetic for explicit confirmation before importing, then surfaces selections pre-toggled in the mark() app. - Use Case: An instructor with 40 student hw01 repositories runs this Skill to pre-grade every question against the rubric, then opens markermd::mark() to review the suggested selections and private notes before finalizing grades. ## Quick Start Apply the stored rubric to all student repositories in my markermd project at ./hw01 and import the marks for review.
