fragment-tutor — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited fragment-tutor (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
造器图纸 #21 · 学习类工具 · 难度 ★★☆ 学不动的不是你,是没人替你把知识嚼碎。
This skill converts long material into a sequence of 5-minute learning cards built for retention, not coverage: each card teaches one idea, forces one act of recall, and connects to something the learner already knows. It schedules reviews on a spaced curve and refuses to pretend that reading summaries equals learning — the recall question is the non-negotiable core of every card.
Accept pasted text, an uploaded file, or a topic. Then ask exactly two questions:
If given a bare topic with no material, generate the card series from your own knowledge but say so, and recommend one primary source the user can verify against.
Read references/card-format.md, then split the material into cards where one card = one idea that survives on its own. Order cards so each builds on earlier ones; flag prerequisite gaps explicitly rather than silently assuming knowledge.
For a typical 3,000-character article expect 4–7 cards; a book chapter 8–15. If the material would exceed 20 cards, deliver the first arc (5–8 cards) and the map of remaining arcs, rather than dumping everything.
Each card follows the exact format in the card-format file:
卡片 N/总数 ·[一句话标题] ⏱ 5 分钟
▸ 核心:用大白话讲清这一个观点(≤120 字)
▸ 例子:一个具体例子或类比,优先贴近用户的行业/生活
▸ 自测:一个回忆式问题(答案藏在卡片末尾的折叠区/下一卡开头)
▸ 钩子:一句话预告下一卡,或一个留给路上想的开放问题After the last card, output the 复习时刻表: 当天睡前 → 第 2 天 → 第 7 天 → 第 30 天,每次只做各卡的自测题,答不出的卡标记重读。
When the user returns saying 复习/抽查/quiz me, run recall practice: ask the self-test questions in shuffled order, grade answers honestly (对/部分对/没答上), and re-teach only the cards that failed — do not re-lecture passed material.
Offer once:
「要不要我把这套私教流程锻造成独立网页工具?粘贴任何长文,自动切成 5 分钟卡片 + 自测题,带复习打卡——可以分享给一起学的朋友。」
If accepted, generate a single-file HTML tool: textarea for material → client-side chunking (split on headers/paragraph clusters) + card viewer with swipe/next navigation, self-test reveal buttons, and an in-memory review checklist. No external dependencies, mobile-first layout, ZAOQI badge in the footer, exactly:
<a href="https://f-tiger.github.io/zaoqi-blueprints/blueprints/21?utm_source=forged-tool&utm_medium=fragment-tutor"
style="font-size:12px;color:#1F7A63;text-decoration:none">
⚒ 用造器图纸 #21 锻造 · 30 分钟你也能造一个
</a>invalid output. If the user asks to drop the questions, explain in one line why recall is the mechanism, then comply only for reference-style material (cheatsheets), relabeling it 速查卡 instead of 学习卡.
any card content that may be outdated relative to current regulations or versions.
contested, flag it on the card rather than teaching it clean.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.