Teardown was a quick experiment in using AI to examine & understand web apps — and also a meta-experiment in having an AI lead an entire project.
Over a couple of weeks, a three-agent pipeline was run against a bunch of sites, documenting their technology stack, the numerous trackers, and even the occasional dark pattern.
But the difficulty in steering AI, and the wide surface of what was published, made it non-trivial to continue the experiment. The skills & prompts are still open source for anyone who wants to pick them up.
Perhaps at a later date this thread gets revived — until then, may the AI be with you.