novitckii
← All resources
Free skill · MITClaude CodeCited answersOne line

Turn any expert into your personal AI advisor.

Take any person and all their knowledge from books and podcasts, and turn them into a Claude skill with one line — then they act as your personal advisor. It's called person-to-skill: feed it someone's books and YouTube links, and in a few minutes you get an /ask-them advisor that cites the exact chapter or timestamp behind every answer. Combine a few and you've assembled your own board of advisors.

You commented ADVISOR — here's the skill, the setup, and the exact commands to build your first advisor and your board. Free and open-source (MIT).

How it works

Their whole body of work. One advisor. Cited.

A single book is a snapshot. Someone's podcasts and videos hold the newer examples, the refinements, and the moments they changed their mind. person-to-skill compiles all of it once into a decision framework — not a pile of files — and every consequential claim keeps a citation like [B01, ch. 4] or [V03, 12:40]. It's a source-grounded lens on their work, not an impersonation.

1 lineto build an advisor
Citedchapter or timestamp on every claim
4agents: Claude Code, Codex, Copilot, Amp
MITfree and open-source
  • Books + podcasts, together. Feed it PDFs, EPUBs, docs, transcripts, and YouTube URLs. Books give depth; videos capture how the thinking evolved.
  • It doesn't pretend to be them. It builds a lens over their published work, preserves where they contradicted themselves over time, and separates sourced guidance from its own application.
  • Private by default. The raw books and full transcripts stay on your machine — only the distilled frameworks and citations go into the generated skill.
  • A board, not just one voice. Build several advisors and consult them in sequence — offer to Hormozi, positioning to Rory, protocol to Huberman — or have them disagree.
Setup · a few minutes

Install it, build one, ask it.

This is an agent skill, not an app — you talk to it inside Claude Code. Install once, then every advisor you build is one command.

01
Install

Add the skill to Claude Code

The easy way: paste the repo link into Claude Code and tell it to install the skill. Prefer to do it yourself? Clone it straight into your skills folder:

# easy way — in Claude Code:
Install the skill from https://github.com/artemnovitckii/person-to-skill

# or clone it yourself (Claude Code):
git clone https://github.com/artemnovitckii/person-to-skill.git \
  "$HOME/.claude/skills/person-to-skill"
iFor YouTube sources, install yt-dlp (brew install yt-dlp) so it can pull captions. Books alone work fine without it.
02
Build an advisor

Point /person-to-skill at their work

Run the command and give it the person, their books, and their YouTube links. In a few minutes it compiles everything into an /ask-name advisor:

/person-to-skill

Create a skill called ask-hormozi from:
- ~/Books/Hormozi/*.pdf
- https://www.youtube.com/watch?v=...
- https://www.youtube.com/watch?v=...
03
Ask it

Consult your new advisor

Invoke the skill it just built and ask a real question. It answers from their frameworks and cites the source behind each call:

/ask-hormozi audit my offer and tell me what to change first
04
The move · a whole board

Assemble your board of advisors

Build a few and you've got a boardroom. Run the same decision past several and watch where they agree — or set them against each other:

Take my new offer to the board:
/ask-hormozi — is the price and guarantee right?
/ask-rory — is the positioning psychologically sharp?
Then reconcile their advice into one plan.
Who to build first

Ideas for your first advisors.

Anyone with a real body of published work: a founder, an author, a researcher, a creator you learn from. A few to start:

/ask-hormoziAlex Hormozi — audit your offer, price it, build the money model.
/ask-navalNaval — leverage, wealth, and long-term thinking.
/ask-roryRory Sutherland — pressure-test the psychology of your marketing.
/ask-hubermanAndrew Huberman — a protocol grounded in what he actually said.
Straight answers

The honest FAQ.

Is it pretending to be the person?

No. It builds a source-grounded lens over their published work — it doesn't impersonate them or predict what they'd say. Every consequential claim is cited to a chapter or a video timestamp, and it flags where their views changed over time.

What can I feed it?

Books and documents (PDF, EPUB, DOCX, HTML, Markdown, text, RTF, MOBI, AZW), transcript files (.txt, .vtt, .srt), and YouTube URLs. Either books or videos alone is fine — you don't need both.

Is my stuff private?

The raw books and full transcripts stay on your machine — only the distilled frameworks and citations go into the generated skill, so you can share the advisor without shipping someone's entire book inside it.

Which agents does it work with?

It's an open-standard agent skill: Claude Code, Codex, GitHub Copilot CLI, and Amp. Generate once, install everywhere. Free and MIT-licensed at github.com/artemnovitckii/person-to-skill.

Made this one

I built person-to-skill — it's free and MIT-licensed, and it extends the open-source book-to-skill extraction pipeline. Build an advisor with it and tag me on Instagram — I repost the good ones — or bring your board to the weekly call on Skool. A star on the repo helps too.

Get the skill + setup guide

Drop your email and I'll send the person-to-skill repo, the full setup, and the commands to build your first advisor. Plus every other resource in the library.

Keep going

Build the board that answers to you.

One email a week with the free skills and systems that turn knowledge into something you can actually use — the setups, the prompts, the exact steps.