CI/CD

A developer reviewing a GitHub outage dashboard on a laptop, illustrating a code hosting single point of failure
Cloud Services

When the Code Host Becomes the Single Point of Failure

For eighteen years, Mitchell Hashimoto opened GitHub every single day. He built his career practically inside it, joked as a 20-year-old that maybe the company would hire him if his open-source project was good enough, and treated the platform the way most of us treat electricity — a utility so reliable it disappeared into the background. Then, in April 2026, he announced he was moving his terminal emulator project, Ghostty, off GitHub entirely. “I want to ship software and it doesn’t want me to ship software,” he wrote. That sentence, more than any ideological complaint, is the real story here — and it’s a story about infrastructure, not politics.

Dashboard showing an LLM release gate with drift, shadow traffic, and canary checks for the focus_keyword
Cybersecurity

When “All Tests Passed” Still Means You Shipped a Broken AI

A Friday deploy can look flawless: every unit test green, every integration check passing, the dashboard practically applauding itself. Then Monday morning arrives, and the system has been quietly telling customers the wrong price all weekend. Nothing crashed. Nothing errored. It just got worse in a way no traditional test was built to notice.

Scroll to Top