Youtube second brain

Skill mike623/agent-skills/skills/youtube-second-brain

Install

npx -y skills add mike623/agent-skills --skill youtube-second-brain

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when turning YouTube videos into durable second-brain knowledge: transcripts, summaries, claims, tags, Obsidian notes, Raindrop metadata, and NotebookLM-ready source packs.

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

4.8 KB, 971 tokens by cl100k_base, as published. Nobody here has run it

YouTube Second Brain

Use this skill when a YouTube video should become reusable knowledge rather than just a saved link. The workflow treats YouTube as a source document: capture, transcript, summary, extraction, linking, and optional deep analysis.

Core principle

YouTube videos become second-brain material only after they are converted into structured, source-grounded notes.

Do not infer video content from title or thumbnail alone. Fetch a transcript when possible. If only metadata is available, state that clearly.

Input sources

A video may arrive from:

a direct YouTube URL;
a Raindrop bookmark;
a playlist;
a browser tab;
a user note such as “this is worth saving”.

Processing levels

Level	Meaning	Action
L0	Save only	Keep bookmark, light tags.
L1	Summary	Transcript summary + useful tags.
L2	Knowledge note	Full note with key ideas, claims, actions, links, and transcript.
L3	Deep research	Mark as NotebookLM/deep-analysis candidate.
L4	Operationalize	Convert into checklist, task, project plan, or reusable skill.

Default to L1/L2 for useful videos. Use L3/L4 only when the video contains reusable strategy, workflows, decisions, or implementation ideas.

Workflow

Capture metadata
- URL, title, channel, date if available, saved date, tags, bookmark ID if present.
Fetch transcript
- Prefer official captions or captured transcript text.
- If transcript unavailable, use public metadata/description only and mark transcript-unavailable or manual-review.
- Do not claim to have watched visual content unless actual video analysis was performed.
Extract knowledge objects
- TL;DR.
- Key ideas.
- Claims and assumptions.
- Tools/products/people mentioned.
- Workflows or frameworks.
- Practical takeaways.
- Open questions.
- Follow-up actions.
- External links from the description/transcript when accessible.
Write a durable note
- Use Markdown with YAML/frontmatter.
- Include source link and transcript provenance.
- Preserve uncertainty and extraction limits.
- Add topic links/tags for later retrieval.
Update capture system when available
- Rename bookmark to a concise searchable title.
- Add a short description explaining why it matters.
- Add tags without removing user tags unless explicitly instructed.
- Mark state such as transcript-extracted, obsidian-note, notebooklm-candidate, or manual-review.
Decide NotebookLM readiness
- Good candidates: dense conceptual videos, multi-source research, product/tool comparisons, strategy, workflows, or claims to challenge.
- Poor candidates: shallow news, entertainment-only clips, duplicate content, or videos with no transcript and little metadata.

Suggested note schema

---
type: youtube-note
source: youtube
url: <url>
channel: <channel-if-known>
saved_date: <date-if-known>
processed_at: <timestamp>
tags:
  - youtube
  - research
status: processed
notebooklm: false
---

# <Searchable video title>

## Source

- Video: <url>
- Channel: <channel>
- Transcript: <official captions / auto captions / captured by bookmark tool / unavailable>

## TL;DR

<short summary>

## Key ideas

- ...

## Claims and assumptions

- Claim: ...
  - Evidence in transcript: ...
  - Verification needed: ...

## Tools / links / people mentioned

- ...

## Practical takeaways

- ...

## Questions for deeper analysis

- ...

## Follow-up actions

- [ ] ...

## Transcript / excerpts

<transcript or relevant excerpts>

NotebookLM source-pack guidance

When preparing for NotebookLM, include:

original video URL;
transcript or transcript note;
summary;
key questions to ask;
related articles/videos/notes;
extraction limitations.

Useful NotebookLM prompts:

Extract the core argument and supporting claims from this source.

Compare this video against the other sources in this notebook. Where do they agree or contradict?

Turn this into an implementation memo with concrete next actions.

Public-safety rules

Do not bypass paywalls, login walls, private videos, or gated content.
Do not fabricate transcript, comments, engagement, or visual details.
Do not commit private bookmark IDs, local vault paths, or personal notes into public skill repos.
Use placeholders in examples.