// michael_bass — chicago · security · ai systems
one keystroke too deep.
welcome inside.
SCROLL
This is not a portfolio — it's a workshop with the lights on.
Everything below runs, ships, or downloads. No slides, no mockups, no demos pretending to work. Two of these tools execute in your browser this second — and nothing you type into them ever leaves it.
On the bench
Live tools. They run entirely client-side — open devtools and watch the network tab stay empty.
LINK X-RAY
Paste a link before you click it. See the anatomy of the URL, what domain it actually goes to, and every structural deception trick it's using — homoglyphs, brand-spoof subdomains, embedded credentials, punycode lookalikes. Verdict in plain English.
OPERATIONAL · runs client-side · collects nothing RUN IT →OSINT CONSOLE
Seventy curated open-source intelligence tools across eleven categories — the working set I actually use, minus the categories I deliberately excluded. The exclusions are the point: professional judgment is knowing what not to touch.
OPERATIONAL · curated by hand · exclusions documented RUN IT →The method
Tooling I built to deliver real client work, published. The engagements stay private — the method is the product.
Workspace offboarding pipeline
WRITEUP IN PROGRESSGoogle Workspace decommissioning at scale: service-account ingest with domain-wide delegation, AI classification of every message, role-based review dashboard, and a runbook that ends with a defensible deletion. The hardest thing I've built.
/cyberteam — security review as a skill
github ↗A CISO-level review framework encoded as an agent skill: seven passes, approval withheld by default. It caught a drive-by localhost RCE in my own release before it shipped. Everything on this site clears it first.
second-brain / second-brain-voice
github ↗A classified, validated personal knowledge corpus — directory-level segmentation, a schema that fails loudly, and a voice layer that's opt-in per note by design.
Client presence audit engine
WRITEUP IN PROGRESSScans a business's digital presence and produces a branded findings report. The interesting part is the honesty rules: anything it can't verify is marked "needs eyes-on" instead of fabricated. An audit tool that refuses to guess.
Disk sanitization lab — NIST SP 800-88
WRITEUP IN PROGRESSMedia sanitization done to the standard, with verification evidence — not "we formatted it."
Storefront & POS builds
DEMOS PENDING REVIEWWhite-label point-of-sale and storefront systems built for real small businesses — a bakery, a market vendor. Sanitized demos link here once they clear security review.
Take it with you
The Loops Build Guide
"One prompt is a guess. A loop is a system."
How to build AI workflows that check their own work. Free PDF — no email, no gate, no form. If it's useful, pass it around. That's the whole idea.
Want something like this built — or the person who builds it?
Employers, contractors, business owners: one door. Tell me what you're trying to do and we'll sort out the rest in conversation.
HIRE ME or write directly: hello@michaelbass.ai
WHAT THIS SITE COLLECTS
Right now: nothing at all. No cookies, no trackers, no fingerprinting, no IP logs kept by me,
and no analytics wired yet. If that changes, it'll be anonymous page-view counts and which tools
get used — never what you typed into them — and this note will say so first. I sell security;
my own site behaving this way is the demonstration.