Tim Henrich Oct 12, 2026 4 min read

tensorlake@0.5.144: A Worm With Valid Provenance, and the Firewall Rule That Stops It

A compromised tensorlake release shipped the Shai-Hulud worm through the project's own CI, with valid build provenance. It was gone from npm within a day. A 72-hour soak time would have kept it off your machines entirely

Read article

Tim Henrich Oct 2, 2026 4 min read

Proxy Settings Are a Suggestion to Your AI Agent

Most attempts to control what a coding agent installs or connects to rely on settings the agent is free to ignore. Here's why that matters and where enforcement has to live instead.

Read article

Tim Henrich Sep 24, 2026 5 min read

What 10.7 million packages taught us about the noise floor

We analyzed 10.7 million package releases across npm, PyPI, crates.io, and Go. Most raise no risk signal at all — here's what the ones that do actually mean.

Read article

Tim Henrich Sep 2, 2026 5 min read

The manual check that saved them won't scale

Last month a software consultancy called Softjourn came a couple of clicks away from installing malware into a client project. An AI agent recommended a malware package to a real engineering team. A manual GitHub check caught it — here's why that check should be automatic and enforced at install time

Read article

Tim Henrich Apr 2, 2026 6 min read

When your security scanner gets infected

Last week, two simultaneous supply chain attacks hit the npm and PyPI ecosystems — UNC1069 (a North Korea-linked group) backdoored axios versions 1.14.1 and 0.30.4, while TeamPCP compromised LiteLLM, Telnyx, and critically, Trivy, a popular security scanner used to detect exactly these kinds of attacks. The Trivy compromise highlights the core weakness of scanner-based security: if your security tool lives in the same ecosystem it's protecting, it's just another attack surface.

Read article

Tim Henrich Feb 19, 2026 2 min read

Securing your Software Dependencies with Rego's Policy as Code

In our previous post Intro to Open Policy Agent for Policy as Code with Regohttps://hextrap.com/blog/introtoopenpolicyagentforpolicyascodewithrego/ we introduce the very basic fundamentals of Open

Read article

Tim Henrich Feb 19, 2026 5 min read

Intro to Open Policy Agent for Policy As Code with Rego

Open Policy Agenthttps://www.openpolicyagent.org/ OPA is a cloudnative policy engine used for enforcing policyascode in major open source projects like Kuberneteshttps://kubernetes.io/,

Read article

Tim Henrich Feb 14, 2026 3 min read

Bakin' some bun into Hextrap

Earlier this year we added support for bun to our list of supported package management tools for Javascript. bun is much more than a package manager for Javascript go

Read article

Tim Henrich Feb 7, 2026 2 min read

Why we added MCP support

Since their inception, Hextrap's firewalls were designed to be used by developers locally and in their CI/CD processes. Now that LLMs have taken over, a new threat has emerged in

Read article

Tim Henrich Jan 13, 2026 3 min read

The Case for Soak Time: Why Waiting 72 Hours Could Save Your Company

Analyzing malicious package detection timelines reveals a simple but effective defense most organizations overlook.

Read article