#Prompts for taking these notes

Copy from this page every time. A prompt you have to reconstruct is a prompt you stop using.

These are written for an AI with video access — YouTube's Ask button (the ✦ next to Share), or Gemini's YouTube integration. Paste the whole block.

#1 · The master prompt

Use this first, on every video.

Give me detailed study notes on this video, structured as:

1. TIMESTAMPED OUTLINE — every section with start-end times.

2. THE METHOD — if a technique is taught, give its exact steps in
   order, using the speaker's own terminology. If he names the steps,
   keep his names AND his ordering.

3. EVERY WORKED EXAMPLE — for each: the exact input he uses, what he
   writes at each step, and the final answer.

4. WHAT HE DRAWS — describe each diagram or tree on screen, node by
   node, with the labels he writes.

5. CODE — any code written on screen, line by line, with the
   explanation he gives for each line.

6. WARNINGS — every "don't do this", "people get this wrong", or
   common-mistake remark, with its timestamp.

7. DISTINCTIVE PHRASES — any phrase or mnemonic he repeats.

8. CONFIDENCE — mark anything you are unsure about, and say which
   parts of the video were unclear to you.

Section 8 is the one that earns its place. Without it these tools state everything with equal confidence and you cannot separate what was transcribed from what was inferred. With it, you know which claims to check.

#2 · Follow-ups

Ask these after the master prompt, when the answer is thin in a particular place.

When two methods exist and you need the decision rule:

What does he say about WHEN to use this method versus the other one?
Give the decision rule as close to his own words as you can.

When the problem list matters:

List every problem he mentions, even in passing, with its timestamp,
and mark whether he SOLVES it here or only names it for a later video.

When a worked example went too fast:

Walk through his worked example again, slower. At each step give me:
the current input, the output built so far, and what the call stack
looks like.

When you want the reasoning, not the result:

Why does he say this approach works? Give his justification, not
just the steps. If he proves or argues anything, reproduce the
argument's structure.

When the video is long and you want to triage:

Which 3 minutes of this video carry the actual technique, and which
parts are motivation, recap, or channel promotion? Give timestamps.

#3 · For a technique video

Adapt the names to whatever framework is being taught.

He teaches [NAME OF FRAMEWORK]. For each named step:
  - his definition, in his words
  - the question he says to ask yourself at that step
  - the ORDER he says to answer the steps in
  - the example he uses to demonstrate it

Also: does he say anything about what NOT to do while applying it?

The ordering question is worth asking every time. Frameworks are usually presented as a list, and the order you apply them in is often different from the order they are named in — and that difference is the part that makes the method usable.

#4 · For a problem video

For the problem solved in this video:
  - the exact problem statement he gives
  - the brute-force or naive approach, if he mentions one
  - the recursion tree or diagram he draws, described fully
  - the base case, and HOW he decides what it should be
  - the final code, line by line
  - the complexity, if he states it
  - any edge case or gotcha he calls out

What does he say makes this problem different from the previous one
in the series? What is the new idea being introduced?

The second question is the valuable one. A series is ordered because each problem adds one thing. Naming that one thing is worth more than the solution, which you could get anywhere.

#5 · The workflow

1. WATCH the video first, at whatever speed suits you.

2. ASK — paste the master prompt into the video's Ask / Gemini panel.
   Follow up with the section-specific prompts where the answer is thin.

3. SAVE — if the answer is long, paste it into a file:
       d:\GITHUB\dsa-handbook\.transcripts\video-NN.txt
   .transcripts/ is gitignored, so working sources never reach the
   published site.

4. SCAFFOLD the page:
       npm run note -- 3 "Hypothesis Induction Base Condition"

5. WRITE it — or hand the summary over and have it drafted, then
   correct it against what you actually saw.

6. PUBLISH:
       npm run publish -- "notes: video 3"

#6 · Two rules for the notes themselves

Mark your sources, and mark your confidence. Every page in this section opens with what it was built from and what has not been verified. That header is not ceremony — it is what lets you trust the page in three weeks when you cannot remember which parts you checked.

Do not delete a claim just because one source is weak. It has already happened once here: a problem was correctly listed, then removed as "unconfirmed" on thin evidence, then restored when a better source arrived. The fix for a bad source is a better source, not a confident deletion. Mark it uncertain and move on.

#7 · What these notes are for

They are a revision artefact, not a replacement for the videos. The material belongs to whoever made it; these pages exist so the ideas can be re-derived quickly later, and every one links back to the original.

If a page here is good enough that you would skip the video, it is wrong — the video is where the understanding comes from, and the page is only where it is stored.