ed

Educational learning skill for ML graduate students — create study notes, pedagogical guides, learning plans, and quizzes using LLMs as a learning tool.

Workflows

new notes -- create Feynman-style study notes

Generate graduate-level notes from lecture slides + deep web research. Notes use Feynman-style explanations, mermaid diagrams, ASCII diagrams, worked examples, sample Q&A, and clickable references.

/ed new notes "attention mechanisms" --resources slides/week3.pdf papers/vaswani.pdf
/ed new notes "backpropagation" --resources slides/lec2.pdf --level undergrad

Missing resources are a hard blocker — the skill will stop and ask you to provide them.

Output: notes/{slug}/notes.md

new guide -- create a pedagogical assignment guide

Break a project or assignment into pedagogical phases with hint-controlled disclosure. Never provides complete solutions — only learning objectives, hints, and approach sketches.

/ed new guide "implement a transformer from scratch" --vim

Flags: --vim appends editor keyboard hints to each step.

Output: guides/{slug}/guide.md

new plan -- create a week-by-week learning plan

Generate a structured study schedule at graduate level. Follows a syllabus if provided; otherwise finds resources via deep research.

/ed new plan "deep learning fundamentals" --weeks 10
/ed new plan "Stanford CS231n" --syllabus syllabus.pdf

Output: plans/{slug}/plan.md

update -- apply annotations

Scan ed artifacts for > (minor) and >> (major) annotations added inline, and apply them in-place.

/ed update notes/attention-mechanisms/notes.md
/ed update    # scan all ed artifacts in cwd

assist -- interactive guided walkthrough

Step-by-step walkthrough of a guide. AskUserQuestion controls how much of each hint is revealed. Supports --vim for keyboard hints.

/ed assist guides/transformer/guide.md --vim

At each step, choose: Got it / Hint / Full approach / Skip / End session.

quiz -- generate or administer a quiz

Create a quiz from notes or slides. Answers go in a separate file.

/ed quiz "attention mechanisms" --source notes/attention-mechanisms/notes.md --questions 15
/ed quiz --interactive quizzes/attention-mechanisms/quiz.md

Output:

• quizzes/{slug}/quiz.md — questions only
• quizzes/{slug}/answers.md — answers only

Output Directory Structure

notes/{slug}/
  notes.md             # Feynman-style notes + diagrams + Q&A

guides/{slug}/
  guide.md             # pedagogical guide with hint blocks

plans/{slug}/
  plan.md              # week-by-week learning plan

quizzes/{slug}/
  quiz.md              # questions only
  answers.md           # answers only (keep closed during study!)

Annotation Format

Add > or >> anywhere in an ed artifact to annotate it for later refinement:

## Attention Mechanism

> clarify the softmax temperature here
>> add a worked example showing the full attention computation for a 3-token sequence

Then run /ed update to consume the annotations and apply them.