tracked my coding sessions for a week. only actually wrote code 51% of the time.
built this as a vs code extension - tracks lines changed, languages used, debug time. intentionally session-based (not always-on) because i wanted to measure my "deep work" blocks specifically.
tech: vs code extension api, next.js, supabase realtime, typescript.
biggest surprise: my 9pm "peak productivity" sessions were 23% coding. morning sessions i thought were slow: 67% active.
happy to answer questions about implementation or the harsh reality of measuring actual productivity.