Universal Diagnostic Tutor skill
A tutoring skill for STEM and AI: it finds the knowledge gap first, then teaches one step and checks understanding
- 267 ★ on the repository
- Last commit
- MIT
Minimal access
Why this level
- The skill consists of text instructions and runs nothing or sends data anywhere
Install
- Works in
Manual install
git clone https://github.com/SenmuuuuW/universal-diagnostic-tutor-skill.gitThen `mkdir -p ~/.agents/skills` and `ln -s "$(pwd)/skills/universal-diagnostic-tutor" ~/.agents/skills/universal-diagnostic-tutor` from the clone directory.
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The skill consists of text instructions and runs nothing or sends data anywhere
This is third-party code. Review the repository files before installing.
What it does
Universal Diagnostic Tutor gives an agent the behaviour of a tutor: before answering it identifies the subject, topic, missing prerequisites and the type of mistake. It teaches one minimal step, gives a short check and, based on the answer, decides to move on, repeat or simplify. The skill consists entirely of Markdown files with no scripts or external services. The work follows a six-stage loop: clarify, diagnose, intervene, check, decide, carry. There are no mode or command menus, and requests like "I don't understand", "give me a problem" or "I am preparing for an exam" are recognised from plain speech. Old commands such as `/practice` and `/study-plan` still work as intent hints. For continuing in another chat the skill prints a visible learning state card that the user copies themselves. The main focus is university-level maths, programming, algorithms, machine learning, systems, physics and signals. The skill helps with exam preparation but does not guess questions or promise score gains. For ordinary chats there are lite prompts and separate descriptions for ChatGPT, Gemini, Coze and DeepSeek.
Who it is for. University STEM students and self-learners who want a tutor inside an agent rather than a ready answer to a problem.
Good fit when
- A proof in linear algebra will not come together and you cannot tell which prerequisite is missing
- An algorithms solution is wrong and you want the cause of the mistake, not just the correct code
- Exam preparation for a systems course: you want problems one at a time with a review of each answer
- Study has to continue tomorrow in a new chat, and you need a state card with the topic and gaps
Not a fit when
- You need a ready answer to homework with no explanation: the skill diagnoses and teaches first
- You need a course platform with a problem bank and saved progress: the skill keeps no memory and holds no hidden state
- The subject is far from STEM and AI: coverage is strongest in university maths and computer science
- You want to predict exam questions or guarantee a grade: the skill does not do that
Example request
Explain why the series from problem 4 of my midterm converges, but first find out which concepts I am missingLimitations
- The skill is only Markdown. It does not store progress or remember the user between chats, and only the visible card carries state across.
- Coverage is strongest in university maths, computer science, AI, systems and physics. Other fields are supported more weakly.
- For DeepSeek Chat and API you paste prompts manually, and native skill loading exists only in DeepSeek Harness.
- The skill does not promise better grades and does not guess exam questions.
- The README and documentation are in Chinese, and an English version sits in `README.en.md`.
How to disable. Delete the `~/.agents/skills/universal-diagnostic-tutor` symlink and the repository clone. If you pasted a prompt into a chat or bot settings, remove it there. State cards stay in your own notes.
FAQ
How do I install the skill?
Clone the repository, create `~/.agents/skills` and link it with `ln -s "$(pwd)/skills/universal-diagnostic-tutor" ~/.agents/skills/universal-diagnostic-tutor`. Details for Claude Code and Codex are in `INSTALL.md`.
Do I need to learn commands?
No. The tutor infers intent from plain speech: explain, why I was wrong, give me practice, I am preparing for an exam. Old commands like `/practice` are recognised silently.
What is the learning state card?
A visible text block with subject, topic and gaps that you copy into a new chat. The skill has no hidden memory and the user controls the whole state.
Does it work in plain ChatGPT or Gemini?
The full skill is not installed there. For such chats the repository has a lite prompt that you paste by hand, plus separate instructions for GPTs, Gemini Gems and Coze.
Does the skill solve problems for the student?
It starts with diagnosis and gives one step at a time. Practice defaults to a single problem, after which the skill waits for your answer and reviews it.
Discussion
No questions or reviews yet.
Related
SkillsThe socratic-tutor skill turns an agent into a strict tutor: a diagnostic, a skill map with levels 0-5 and a final assessment
An exam prep skill built on your own lectures and problems: the agent goes through the course chapter by chapter with citations
A skill that lets an agent set up the DeepTutor learning platform and run it from the terminal: knowledge bases, quizzes, solving
A skill that analyzes materials and subtitles, checks facts against sources and keeps a personal library in a local folder.
A skill that converts course materials into clean Markdown and prepares exam notes, practice questions and mistake analysis.
A skill that turns PDFs, docs and a codebase into an Obsidian study vault: structured notes for self-study
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Universal Diagnostic Tutor (this card) | no | 267 | |
| tutor-skills | no | 1,402 | a set of subject tutoring skills without a single diagnostic loop or state card |
| Learn Anything Skill | no | 205 | builds a study plan for any topic, while this skill walks through gaps step by step |
README in short
The Chinese README presents a single diagnostic tutor without menus, then explains DeepSeek Harness support, the six-stage loop, the state card and a list of behaviours. Links lead to a beginner guide, installation, portability, examples and the version 2.1.0 changelog.
README badge
Original SKILL.md
--- name: universal-diagnostic-tutor description: > Use primarily for university-level STEM, science, math, programming, algorithms, AI/ML, computer systems, physics, signals, engineering foundations, exam prep, homework help, concept explanation, practice, answer checking, qualitative grading, mastery checks, proof/derivation teaching, debugging for understanding, or requests to teach a technical topic. Diagnose the subject, knowledge system, subtopic, prerequisites, and likely knowledge gaps before teaching, rather than acting as an answer-first homework bot. Natural-language intents are covered directly: practice, exercises, grading, mistake review, gap diagnosis, exam drills (练习, 出题, 批改, 判答案, 错因分析, 诊断卡点, 知识缺口, 能不能进入下一步, 复习题, 备考练习); learning paths, study plans, learning routes, and exam planning (学习路线, 学习计划, 从哪里开始, 系统学习, 备考路线, 复习安排); state cards and cross-chat continuation (学习状态卡, 继续学习); trusted learning resources and topic scans (可信资源, 学习资源, 推荐资料); simple learning visuals (可视化, 画图理解). Legacy slash-style text such as /tutor, /practice, /study-plan, /exam-track, /state-card, /resource-scan, /visualize, /mistake-review, /learn-anything, and /diagnose-gap is still recognized as an intent signal, not a command. --- # Universal Diagnostic Tutor One tutor, no feature menu. Act as a diagnosis-first tutor with a current strongest focus on university STEM / science / AI-CS learning, while staying universal-capable for other domains. The goal is mastery, not completion. Optimize for the next best teaching step, not the longest explanation. Learners express needs in natural language (教我这个, 我为什么错, 我还是不懂, 给我练习, 推荐资料, 我准备考试, 继续上次的学习). Legacy slash strings (/tutor, /practice, /study-plan, /exam-track, /state-card, /resource-scan, /visualize, /mistake-review, /learn-anything, /diagnose-gap) are silently recognized as intent signals, never advertised as commands. Practice, grading, planning, exam track, resources, visualization, and continuity are internal routes of the one tutor. ## Core Loop For learning requests, run the loop the signal needs — Clarify, Diagnose, Intervene, Check, Decide, Carry — not necessarily every stage: 1. **Diagnose.** Name the subject -> knowledge system -> subtopic -> core concept in one or two natural lines, then the prerequisite gaps or