case-answer-review

Grades Korean law case-study answers and generates review packets from textbook indexes and RAG results.

Updated Jun 15, 2026
One-click install
npx skills add https://github.com/kmjy98-sketch/khulaw --skill case-answer-review-kmjy98-sketch
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: case-answer-review
Source: https://github.com/kmjy98-sketch/khulaw/tree/main/.agent/skills/case-answer-review
Command: npx skills add https://github.com/kmjy98-sketch/khulaw --skill case-answer-review-kmjy98-sketch

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires pypdf, and includes scripts (resource) components.

What problem does it solve? Manually grading Korean law case-study (사례형) answers and assembling related study materials is slow and inconsistent. This Skill automates provisional grading, consistency checks against model answers and explanations, and collection of related issues, statutes, precedents, and practice problems into a single review report. ## Core Features & Use Cases - Provisional Grading with Rubric: Scores answers on issue spotting, keyword match, structure, conclusion, and fact-to-law application using a weighted rubric, marking items as pending when source material is missing. - Consistency Verification: Cross-checks the answer against question sheets, model answers, explanations, and textbook RAG chunks, citing exact evidence locations for every deduction. - Study Packet Generation: Auto-matches case material bundles by lecture, converts multiple-choice items into O/X drills, analyzes weak points, and proposes spaced-repetition and Anki card follow-ups. - Use Case: A law student submits a civil law answer draft on capacity to act; the Skill produces a Markdown review report with matched issues, grading, missing points with source citations, related past exam problems, and suggested review items saved to case_answer_packet.json. ## Quick Start Run the render_case_answer_review.py script with the subject, topic, and answer text to generate a provisional grading report and study packet.

Frequently Asked Questions about case-answer-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I grade a Korean law case-study answer automatically?▼

Run render_case_answer_review.py with the subject, topic, and answer text or file path. The script loads problem indexes and learning state, applies a weighted rubric for issue spotting, keywords, structure, conclusion, and application, then writes a Markdown report and a case_answer_packet.json packet.

What input file formats does the case answer review script support?▼

The script accepts plain text, Markdown, PDF, and DOCX answer files. PDF text is extracted with pypdf, DOCX paragraphs are parsed from the document XML, and inline text can also be passed directly via the --text argument.

What happens when the model answer or explanation is missing?▼

Grading falls back to a provisional evaluation only, and affected rubric items are marked as pending rather than scored. If a model answer exists without explanation or textbook RAG support, the report flags it as textbook/explanation verification deferred.

Does the grading script automatically record scores to the spaced repetition log?▼

No, the script never writes to srs_log.json silently. It only proposes review items and Anki card candidates based on the weakness rules, and actual recording happens only when the user confirms and invokes the spaced-repetition skill.

Why are some multiple-choice questions not converted to O/X drills?▼

Only direct O/X questions convert immediately. Numbered-choice questions convert per-choice only when choice texts are stored in the index or parseable from the question text; otherwise they are listed as pending due to insufficient choice source text.