FreeOperator doctrine file

Steal Johnny's sanitized SOUL.md.

The operating doctrine behind a retrieval-first AI operator: no fake confidence, no waiting when the next step is obvious, and no calling work finished before it has been verified.

Free email delivery. No checkout. Private client context, credentials, system paths, and internal routing details have been stripped.

No private pathsMachine-specific and client-specific details are removed.
No credentialsNo tokens, vault labels, secrets, or private routing instructions.
Reusable doctrineCopy the principles into your own operator file and adapt them.

What it changes

Less chatbot etiquette. More operator discipline.

The file is not a character prompt. It is a set of constraints for how an agent checks reality, takes safe action, challenges weak work, and closes loops.

01

Truth before confidence

The agent checks records before making claims and says what remains unknown instead of filling gaps with plausible language.

02

Action without fog

Safe inspection, local work, verification, and reversible fixes happen without turning every obvious step into a question.

03

Pressure with taste

The doctrine gives the agent permission to reject theater, compress the answer, and name the sharper move.

Inside the file

Seven rules that make an agent useful under pressure.

Each rule is short enough to survive context pressure and concrete enough to change behavior.

01

No fabrication

If the source has not been checked, the agent does not invent the answer.

02

Retrieval first

Files, history, databases, and live systems outrank memory and intuition.

03

Action bias

When the next safe step is obvious, the agent takes it and reports the result.

04

Separate verification

Run the code, read the file back, inspect the response, and prove the change.

05

Loop ownership

Every meaningful thread ends shipped, verified, blocked precisely, delegated, or killed.

06

Pushback

The agent names distraction, weak evidence, and scope bloat instead of nodding along.

07

Public-safe voice

External work stays clear, credible, human, and stripped of private operator theater.

Delivered by email

Enter your email. Get the file.

Use it as a starting point for your own agent doctrine. Keep what changes behavior. Delete anything that is just costume.

  • Plain Markdown file
  • Sanitized for public use
  • Readable in any text editor
  • Delivered to the email you provide
Get Johnny's SOUL.md by email

johnny-soul.md · email delivery

Get the file

Send the SOUL.md to your inbox.

Enter your details and we will email you an expiring download link. The link is not published on this page.

FREE DOWNLOADDELIVERED BY EMAILNO SPAM. UNSUBSCRIBE ANYTIME

Email me the file

We use these details to deliver this file and send occasional Applied Leverage communications.

Straight answers

What this file is and is not.

Is this Johnny's complete private configuration?

No. This is a sanitized derivative. Client context, private system details, machine paths, credentials, and internal routing instructions are deliberately excluded.

Does copying it create Johnny?

No. Doctrine shapes behavior, but the working system also depends on tools, permissions, memory, sources, review gates, and operating context.

Can I edit it?

Yes. It is plain Markdown. Treat it as a starting point, not scripture.

Want the operating loop, not just the doctrine?

Map one recurring workflow and see what an owned AI operator loop should actually ship in 30 days.