Diagnoses and fixes "smashed paragraphs" on this site — bullet points, numbered steps, or other structured lines that render as one run-on paragraph (joined by spaces/dashes) instead of separate list items or blocks. Use this whenever the user says content looks "smashed", "run together", "merged", "not spaced out", or points at a lecture/notes/DAWRAT page where text that should clearly be separate lines/bullets is showing up as one dense paragraph. This is almost always a parser bug (in parser/core/collectors.js or parser/blocks/handlers.js), not a content-authoring problem — fix it at the parser level so it's solved for every lecture at once, not by manually reformatting one file.
Answers and organizes an already-extracted past-exam MCQ file (دورات/DAWRAT) for a subject in this repo — fills in the blank الإجابة الصحيحة/التعليل fields by matching each question against that subject's own lecture content (mainly its "النص الأصلي يقول" quoted paragraphs), groups the questions by which lecture their answer came from, and writes subjects/<year-N>/<subject-id>/DAWRAT/exams.md + manifest.json. Use this whenever the user asks to answer, fill in, combine, or organize a DAWRAT/دورات past-exam file, or mentions matching exam questions to lecture answers — even if they don't name the skill or say "DAWRAT" explicitly. This picks up AFTER templates/prompt-past-exam-mcq.md's extraction step (which only pulls raw questions out of a PDF with blank answers) — don't use this skill to do that extraction itself, and don't use it to touch any file outside one subject's DAWRAT/ folder.