菜单
Store, validate, review, confirm, and query Peter Zhou mistake-bank records. Use for RawMistakeRecord persistence, SourceDocument status, is_wrong review, id allocation, crop refs, and script-owned JSON writes.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Generate and print correction practice, then grade returned answer sheets. Use for CorrectionPaper PDFs, PDF/PNG printing, original redo, generated similar practice, PracticeQuestionRecord lifecycle, returned scan/photo/PDF grading, and mastery counters.
Generate structured wrong-question explanations and rich local teaching pages. Use for ExplanationArtifact caching, crop re-interpretation, Xiaohei-style teaching illustration plans, verified optional video recommendations, and TeachAny-style local ExplanationPage output.
Create reusable graph-grounded system knowledge lessons for recurring weak topics. Use for learning-unit selection, agent teaching drafts, controlled explanatory animation, linked practice, and archived HTML courseware.
Compute Peter Zhou learning overview reports. Use for total wrong counts, subject breakdown, mastery progress, repeat-correction markers, weak knowledge tags, correction candidates, and recent source/correction activity.
Recognize wrong questions from scanned/photo/PDF papers using model-native vision. Use for source manifests, page order reasoning, teacher-mark interpretation, unmarked-paper grading, and RecognitionResult creation.
Student-facing wrong-question super-teacher navigator for scanned/photo/PDF student papers. Use when the user starts Peter Zhou, asks for learning status or next actions, recognizes wrong questions, stores a mistake bank, explains mistakes, generates or prints correction practice, grades returned papers, learns a knowledge point, or reports a Peter Zhou problem from an installed user host.
| name | peter-zhou-mistake-bank |
| description | Store, validate, review, confirm, and query Peter Zhou mistake-bank records. Use for RawMistakeRecord persistence, SourceDocument status, is_wrong review, id allocation, crop refs, and script-owned JSON writes. |
Use this subskill when a recognized result must become canonical local data.
Core contract:
SourceDocument state separate from subject mistake records.is_wrong=false records for audit, but exclude them from final stats and practice by default.scripts/allocate_id.py and scripts/json_store.py; see references/schema.md.scripts/intake_recognition.py ingest, present the returned review_report as the next skill step for user review. Focus the user on needs_review=true, low-confidence answers, ambiguous teacher marks, and crop screenshots. If the user does not review, leave the source status as processed; only apply-review --confirm moves it to confirmed.scripts/intake_recognition.py review-report to regenerate the same derived report, apply-review for user corrections/confirmation, and repair-crops for script-owned screenshot bbox fixes.