Using LLMs in CI/CD for semantic testing of web content

This is sick. Honestly, this solves a huge pain I’ve run into a bunch; knowing a site “works” but having zero clue if it’s actually good or on-brand without someone manually combing through it.

Love how you’ve wrapped all this in stuff devs already use (Jest, Docker, Testcontainers). No weird tooling, no “just trust the LLM” vibes. And keeping the prompts readable-as-tests? Chef’s kiss.

Genuinely feels like the kind of thing we’ll all be doing a year from now and wondering why we didn’t start sooner.