8 read how you think. Protocol ran the work. In v0.6, that split is gone: one instrument sees your thinking and runs the work loop with you — scope, assess, brief, review, adjust.
The proof of the merge is brief.
Ask 8 to brief a model for a real task and it now drafts a usable prompt in the same turn. Inside that draft, it labels the judgment points:
Instead of grinding through empty back-and-forths, you get a working prompt in hand right away — one you can review before you ever send the work to an AI. The point is not to replace your judgment; it is to make the draft inspectable enough for judgment to happen.
8 also watches for a common AI failure mode: agreement that gets stronger while nothing new gets checked. When that happens, it asks for the counterexample before you adopt the answer. That helps keep the work grounded in reality, not just in a language model's praise.
And it runs inside the AI you already use, like Claude, ChatGPT, or Gemini. No new app. No new habit.
8 v0.6 is live at ltwilson.com/8.