Things I build, break, fix, and write about

29 July 2026
Headlamp in-cluster deployment and RBAC boundaries

Headlamp in-cluster deployment only makes sense when the UI belongs inside the cluster, and even then the service account should be painfully narrow. I trust the boundary, not the person using it on a quiet afternoon.

28 July 2026
Marketplace takedown pressure on vendor operations

Marketplace takedowns do more than pull a site offline, they cut through the vendor routines that keep sales moving. In dark web marketplaces, that...

28 July 2026
Brickstorm persistence in Microsoft 365

Brickstorm showed how Microsoft 365 persistence can survive a tidy password reset and a few blocked logins. I keep coming back to the same point, if...

27 July 2026
Restricting Serv-U access before patching

The SolarWinds Serv-U flaw is the sort of bug that turns a routine patch into an ugly afternoon, so I would block the obvious request pattern first...

Latest blog posts you might like

20 July 2026
Weekly Tech Digest | 20 Jul 2026

Stay updated with the latest in tech! This digest covers AI ethics, auto industry shifts, and the impact of politics on technology, exploring today's pressing issues.

19 July 2026
Orthanc upload paths and DICOM heap overflow

Orthanc does not get to be picky, so the upload path is where the trouble starts. A DICOM heap overflow is rarely elegant, it usually begins with one bad length field, one careless allocation, and a...

19 July 2026
CVSS and EPSS for faster triage

EPSS is the bit people miss when they sort by CVSS alone. I use it to catch the vulnerabilities that are actually getting swung at, not the ones that merely look nasty on paper, and that changes the...

18 July 2026
Hardening Salesforce access after token compromise

Hardening Salesforce access after a token compromise is mostly about being blunt before being clever. I want the broken token chain gone, the document paths checked, and the logs in one place, because...

18 July 2026
How .env theft exposes AI service API keys

A stolen .env file is not just a mess to tidy up, it can hand out API keys that keep working long after the box is gone. That is how autonomous attack workflows stay alive, with one credential quietly...

17 July 2026
OCI image trust for Linux host observability

Inspektor Gadget OCI images make deployment easy, but they also move trust to places people rarely examine. I would rather be awkward about signatures, TLS and build inputs now than discover later...

17 July 2026
Object.prototype pollution in Acrobat Reader DC parsing

Adobe Acrobat PDF parsing is one of those areas where a small trust mistake becomes something much nastier. I spent time tracing how polluted prototypes slip into privileged JavaScript, and the path...

16 July 2026
LiteLLM exploit chains at the auth boundary

LiteLLM exploit chains only work because the auth boundary is doing too much, and failing badly when the database wobbles. I trust a proxy less when a bad lookup can quietly turn into proxy-admin...

16 July 2026
Manual validation in AI-assisted pentesting

AI-assisted pentesting is useful until it starts sounding certain, and that is where I get suspicious. The machine can surface a neat story, but manual validation is where you find out whether it...

15 July 2026
Corecrypto proofs against FIPS 203 and FIPS 204

Apple’s corecrypto formal verification work is interesting because it is properly dull, in the best sense. It ties a portable C implementation back to FIPS 203 and FIPS 204, then leaves the optimised...