Agent Skills by zhyu
Showing 42 vetted skills indexed across 1 GitHub repositories.
AKCqhzdyhkdse-subject-grading
Extract PDF rubrics, calibrate reference data, and grade HKDSE exam papers.
AKCqhzdyhkdse-subject-evaluation
Evaluate AI-generated HKDSE subject grades against ground-truth labels using ordinal metrics.
AKCqhzdyhkdse-divided-grading
Automate divided-part grading workflows for HKDSE elective exams.
AKCqhzdymodel-service
Configure YAML pipelines to call LLM APIs with environment variables.
AKCqhzdyextraction
Extract structured text data from PDFs or images via YAML-configured pipelines.
AKCqhzdyhkdse-ict-evaluation
Compare AI-generated HKDSE ICT grades to ground-truth levels using rank-based metrics.
AKCqhzdyhkdse-chinese-evaluation
Compute rank-based metrics and generate DOCX reports for HKDSE Chinese AI grading.
AKCqhzdyhkdse-ict-grading
Extract rubrics, calibrate reference data, and grade HKDSE ICT exam responses.
AKCqhzdyhkdse-ict-report
Generate HKDSE ICT class and per-student DOCX reports from AI-graded JSON data.
AKCqhzdyhkdse-chinese-grading
Grade HKDSE Chinese Language papers with rubric-based scoring and per-student JSON output.
AKCqhzdyextraction-v2
Extract structured text from exam PDFs and images into v1-schema JSON.
AKCqhzdysubject-grading-v2
Grade anonymized student answers against rubric-driven schemas into per-question and total scores.
AKCqhzdylingliang-grading
Grade Lingliang primary school math exams via rubric extraction and reference calibration.
AKCqhzdyhkdse-math-evaluation
Compute Spearman ρ and Kendall τ for HKDSE Mathematics AI grading against ground-truth levels.
AKCqhzdylingliang-evaluation
Compare AI-generated Lingliang math grades against ground-truth scores using Spearman ρ, Pearson r, MAE, RMSE, and ±5-point accuracy.
AKCqhzdyhkdse-chinese-report
Generate HKDSE Chinese DOCX reports from AI-graded JSON data.
AKCqhzdyhkdse-math-grading
Grade HKDSE Math Paper 1 papers with rubric extraction and JSON outputs.
AKCqhzdygrading-level-division-reference-calibrated
Align grading boundaries with reference score distributions and emit level_division.json.
AKCqhzdyeval-grading-quality-audit
Audit per-question grading quality by running sub-agents that cross-check evidence against rubrics.
AKCqhzdyorch-subject-pipeline
okay, here's your JSON: {}
AKCqhzdygrading-qa
Inspect grading outputs and summaries for quality issues across directories.
AKCqhzdygrading-guide-render
Render a byte-stable grading guide from validated questions.json and rubrics.json.
AKCqhzdygrading-score-compilation
Aggregate per-student grading JSONs into final_scores.json with per-question statistics.
AKCqhzdyshared-subagent-batching
Coordinate parallel per-unit sub-agents with retries and aggregate validation.