Third-party tutorials about your product are a document your team never reads correctly. You see them as flattering ecosystem content; they're actually usability bug reports with production values. Every step a tutorial slows down to explain is a step your interface failed to explain. Every "make sure you click X first, or it won't work" is a sequence error your design permits. Every workaround taught to thousands of viewers is a missing feature with measured demand.

The tutorial ecosystem is your product's shadow documentation — written by outsiders, shaped entirely by where real users need rescuing. Read as UX research, it's a continuously updated map of exactly where your product fights its users.

deep_search collects the shadow documentation: tutorials and how-to content about your product (and your rivals'), transcript-searched so "tutorials that explain how to work around <limitation>" finds the confession directly.

library_import_youtube_video pins the corpus; deep_context interrogates it like a research repository: "which steps do tutorials consistently slow down on? what warnings do they repeat? which workarounds get taught?" — answers cited to the tutorials. The findings read as a usability audit conducted by your ecosystem: friction ranked by how many independent tutorials address it, each item with the exact clip for the design review.