7 Minute Security

The core actionable insight - install LibreOffice via pip so Claude can render and self-check its own PowerPoint output - is genuinely useful, but it is buried under extended personal tangents (bug bites, winery gig, guitar review, grief) that consume roughly a third of the runtime. The AD attack chain is real practitioner knowledge but is described narratively rather than taught methodically.

“install LibreOffice and a couple other things from uh, Python Pip repository. So I punched all that in and, and uh, then said, uh, okay, why don't you take a look at your early draft and see what needs to be tightened up. And then I'll say it moved like mountains forward in quality”

“I need a couple specific bits of data. I want a PowerShell script that leaks a local admin username and password”

Using Claude to fabricate realistic-looking terminal output with themed fake usernames (Halloween horror characters) as a presentation aid is a genuinely creative workflow hack not commonly discussed. However, the broader thesis - 'AI is good at making slides' - is well-trodden, and the ethical AI hand-wringing is recycled discourse.

“I took a picture like a JPEG of what the screen looks like. I also copied the raw content and I fed it to Claude and I said, you know, give this all a Halloween theme”

“I still felt like kind of the creative director, but I let the visual artist, in this case Claude, do some of that heavy lifting and make an aesthetically pleasing deck that I wouldn't have been able to do with quadruple the hours of time”

This is a solo monologue episode with no guest whatsoever. The host is a working penetration tester with genuine operational experience and describes real attack chains from actual engagements, which lends some practitioner credibility, but the absence of any guest means this dimension cannot score higher.

“abuse a scheduled task while a DA is interactively logged in and fire the task under the DA's context to make them do something high powered like add you to the domain admin group. That's super fun, that's super cool. And you don't need to know the DA's password to uh, pull that off”

“C password values from that MS.14 025 vulnerability”

The episode earns points for naming real tools (Snaffler, Snaffle PI, Rubius, NetExec/NXC, WinRM, ADSI Edit, N8N, LibreOffice, Proxmox, Meraki, OPNsense), real vulnerabilities (MS14-025), and concrete workflow steps; the Halloween-themed demo with named fictional users adds memorability. It loses points because most numbers are casual estimates and no hard metrics, benchmarks, or client outcomes are cited.

“there's, uh, 20 people here who haven't logged in since 2023. And oh, their passwords never expire either. So if I take a look at some zombie data with Snaffle Pie or Snafler”

“C password values from that MS.14 025 vulnerability”

There is no interview, no guest, and no structured Q&A - this is an unscripted solo ramble that drifts freely between AI tool tips, an AD attack narrative, insect-bite urgent care, a winery gig, guitar reviews, and grief over a deceased parent. There are no probing questions, no follow-ups, and no productive tension of any kind.

“Hello, friends, and welcome to the 7 Minute Security Podcast. My name is Brian Johnson and I am your, uh, kind of optimistic, little tired, little thirsty, little itchy. I gotta tell you about that.”

“That's going to be my segue into this personal tangent about itching myself till I bled.”