This is not a personality test and not a clinical claim. It is a behavioral reading of your operator style using only telemetry structure from Thu, Mar 19 through Mon, Mar 23: timing, switching, app/domain gravity, typing distribution, and clipboard behavior. No keystroke content was used to build this profile.
You do not behave like a single-threaded maker trying to protect a monk-like bubble. You behave like a conductor of tools, models, and contexts who can re-enter build mode fast, but pays a real tax when admin and publishing surfaces mix into the same operating field.
You appear to think by steering systems and models, not by drafting in isolation.
Switches happen often, but the return path back into build mode is extremely fast.
You scan broadly for leverage, examples, tools, and opportunities before collapsing back to build.
When a day is meant to be a build day, you can make it very pure.
Your serious output shows up after noon, not in a classic morning-maker pattern.
82.2% of all typed characters in the five full days landed inside Claude, Codex, or OpenWork, with 1,717 paste events into those tools. That is not a person trying to retreat from systems. It is a person using models as thinking surfaces, delegation surfaces, and intermediate memory.
Median return from communication back into build mode was 0.36 minutes. From research back into build it was 0.26 minutes. That suggests you are not fragile in the face of interruption. You can bounce out, collect signal, and snap back.
Across the full five-day window, 84.9% of non-ambient focus stayed in build surfaces. The cleanest execution day was Sat, Mar 21 at 97.2% build-share with 31,246 characters typed. You do not look like someone who benefits from open-ended, mixed-mode days.
The widest exploration day touched 60 distinct browser hosts, but your dominant hosts still clustered around build and AI surfaces. That pattern reads as strategic scanning rather than random doomscrolling: you range widely, then collapse back into a small execution stack.
Only 1,754 typed characters landed in the morning across the full five-day window. Afternoon and evening carried 58,727 and 44,869 characters respectively, with even late night still active at 19,341. The data does not support a fake 9-to-5 self-image.
These are observed foreground stretches after merging same-mode activity. They are useful as state-shape evidence, not exact stopwatch truth.
You look like someone who gets leverage by turning thought into orchestrated external structure fast. You appear comfortable thinking in public with tools, comfortable revising in-flight, and unusually capable of resuming a build thread after stepping out for signal.
The real risk is not distraction in the usual sense. It is over-mixing modes. The same traits that let you steer multiple copilots and surfaces can also blur the line between decisive progress and expensive steering. Your best state is not “less tooling.” It is sharper regime boundaries.
Best evidence for that: 84.9% build-share across the full five-day window, 1,717 AI paste events, and sub-minute median returns back into build mode after comms or research.