When does your human check in?
Analyzes the timing of user follow-up messages within Claude Code sessions. Shows whether humans supervise early, check in mid-session, or review near the end — and how long they trust CC to run solo.
npx cc-checkin
npx cc-checkin --jsonOr open index.html in a browser and drag in .jsonl files.
- Autonomous rate — sessions with zero follow-up messages
- Check-in timing — early (0–33%) / mid (33–67%) / late (67–100%) of session
- Trust duration — time before the human first checks in
- Longest trust runs — sessions with the longest autonomous stretch before any check-in
cc-checkin — When does your human check in?
Sessions analyzed: 162
Session type
autonomous 30% (48) zero follow-ups after initial prompt
interactive 70% (114) human checked in at least once
When do check-ins happen? (939 total check-ins)
early 31% (290) 0–33% into session — supervising startup
mid 26% (246) 33–67% — checking on progress
late 43% (403) 67–100% — reviewing near the end
Trust duration
median 9min — typical autonomous run before human checks in
max 15.8h — longest solo run before first check-in
avg/session 8.2 check-ins in interactive sessions
30% of sessions run completely uninterrupted.
Median trust: 9 minutes autonomous before the human looks in.
MIT
Claude Code Ops Kit ($19) — 16 production hooks + 5 templates + 3 tools. Built from 160+ hours of autonomous operation.