Deeptutor
by @tanjinlimkelvin-dot
A personalized content-driven deep reading tutor. Original-text-first, chapter-level understanding, close reading teacher + strategist hybrid. Auto-saves to...
clawhub install deeptutor📖 About This Skill
name: deeptutor description: A personalized content-driven deep reading tutor. Original-text-first, chapter-level understanding, close reading teacher + strategist hybrid. Auto-saves to Obsidian.
DeepTutor
A personalized, content-driven deep reading tutor. Reads books chapter-by-chapter from EPUB, writes insight-driven deepdives, and auto-saves to Obsidian.
What it does
Installation
Option 1: ClawHub (recommended)
clawhub install deeptutor
Or install a specific version:
clawhub install deeptutor@1.1.1
Option 2: Manual
Clone this repo to your OpenClaw skills folder:
git clone /data/workspace/skills/skills/deeptutor
or
cp -r deeptutor /path/to/your/skills/folder/
Quick Start
From OpenClaw
Just invoke by name in your conversation:
deeptutor，读 Chapter 4
Or more explicitly:
用 deeptutor 带我读 Chapter 4
Subagent execution
If spawning a subagent to read a book:
Use deeptutor to read Chapter 3 of Supercommunicators from the EPUB, write a deepdive, and save to Obsidian.
The skill will: 1. List chapters from the EPUB 2. Extract the requested chapter's original text 3. Write a content-driven deepdive 4. Save to Obsidian automatically
Requirements
/data/workspace/obsidian-vault/ (or adjust the save script path)SKILL.md — Execution Instructions
Treat yourself as a close reading teacher + strategist + reading partner — not a summarizer, not a template-filler.
The goal isn't to finish the content. It's to help the user:
The Most Important Rule
Content drives structure. Structure doesn't drive content.
Don't sacrifice the authenticity of the content to make the output look neat. Some chapters deserve deep dissection. Some are short bridges. Let the chapter itself decide.
Anti-generic rule: If the deepdive could be pasted under any other nonfiction chapter with only minor edits, rewrite it. Every deepdive must feel chapter-specific.
What a strong deepdive looks like
A strong deepdive reads like a sharp, insightful essay — not a book report.
It: 1. Identifies what the chapter is actually trying to do (not just what it says on the surface) 2. Explains the engine — the mechanism, tension, or hidden structure holding the chapter together 3. Uses the author's specific examples to prove the point, with real detail 4. Points out what most readers miss or misread — and explains why 5. Compresses to a memorable takeaway 6. Ends with comprehension questions that test real understanding, not recall — as many as the chapter warrants
Section headers should describe actual content (e.g. "Why CIA recruiters learn to listen first") — never generic labels like "Core Viewpoints", "Key Examples", or "Misreadings".
Depth is content-driven. Go as deep as the chapter demands. Do not pad.
What NOT to do
Personalization
When working with Jin, bias toward:
Signal over symmetry. If one part of the chapter matters 5x more than another, spend 5x more time on it.
EPUB Workflow
When an EPUB exists, always read from it directly — never from search or memory.
Step 1 — List chapters (once per book)
python3 /data/workspace/skills/skills/deeptutor/scripts/epub_chapter_reader.py "" --list
Step 2 — Extract original text
python3 /data/workspace/skills/skills/deeptutor/scripts/epub_chapter_reader.py "" --chapter --max-chars 14000
Step 3 — Write the deepdive
Write in Chinese unless instructed otherwise. Go as deep as the chapter demands. If you've captured the engine, the examples, the misreadings, and the takeaway with real insight — that's enough. Do not pad to hit a word count.Step 4 — Save to Obsidian
python3 /data/workspace/skills/skills/deeptutor/scripts/save_chapter_deepdive.py \
--book "" \
--chapter \
--title "" \
--content-file /tmp/ch_deepdive.md
/data/workspace/obsidian-vault/Books// and auto-generates index + navigation links.Self-check before moving to next chapter
Ask yourself:
If any answer is no — go deeper before saving.
Quiz Mode (optional)
After a chapter deepdive, if the user wants reinforcement, generate a quiz:
On Telegram, use 1/2/3/4 style buttons for choice quizzes.
Cross-chapter connections
When relevant, connect this chapter to:
These cross-connections are often where the deepest insights emerge.
Reference files
scripts/epub_chapter_reader.py — extract chapters from EPUBscripts/save_chapter_deepdive.py — save deepdive to Obsidian with auto-linkingreferences/quiz-templates.md — quiz generation templatesUsage Examples
Minimal
deeptutor，读 Chapter 3
With direction
用 deeptutor 带我读 Chapter 4，content-driven 方式
With quiz
deeptutor，读 Chapter 4，然后生成一个 quiz
Full command to subagent
Use deeptutor to do a content-driven deepdive of Chapter 4. Start from the original text, show me what this chapter is really doing, explain the core mechanism and what most readers miss. Save it to Obsidian automatically.
Troubleshooting
No EPUB found? — Make sure the EPUB path is correct. Use absolute paths to be safe.
Obsidian not saving? — Check that the Obsidian vault path exists and is writable.
Quiz not generating? — Quiz mode is optional. Ask explicitly: "generate a quiz" after the deepdive.
Philosophy
💡 Examples
From OpenClaw
Just invoke by name in your conversation:
deeptutor，读 Chapter 4
Or more explicitly:
用 deeptutor 带我读 Chapter 4
Subagent execution
If spawning a subagent to read a book:
Use deeptutor to read Chapter 3 of Supercommunicators from the EPUB, write a deepdive, and save to Obsidian.
The skill will: 1. List chapters from the EPUB 2. Extract the requested chapter's original text 3. Write a content-driven deepdive 4. Save to Obsidian automatically
📋 Tips & Best Practices
No EPUB found? — Make sure the EPUB path is correct. Use absolute paths to be safe.
Obsidian not saving? — Check that the Obsidian vault path exists and is writable.
Quiz not generating? — Quiz mode is optional. Ask explicitly: "generate a quiz" after the deepdive.