Launched this week
Chit reads the Claude Code transcripts already on your disk and prints the day back as a receipt, grouped by project and ready to paste into a standup. A 777 KB Mac app. No Hammerspoon, no Python, no account, and no network code in the binary at all.
The receipt format is a clever way to solve the end of day problem. I have definitely finished a productive session and then struggled to remember exactly what i accomplished.
Notchling
@charles_brewer90 That gap is exactly why I built it — the more productive the session, the less of it I could remember afterwards. Thank you. Curious what you'd paste it into: a standup, a timesheet, or just your own notes?
Build Check
Notchling
@german_merlo1 Thank you, means a lot!
Interesting
Notchling
@madalina_barbu Thanks for checking it out!
Dial
"no network code in the binary" is the detail that actually gets something like this installed instead of just starred and forgotten. we run a few parallel Claude Code sessions across different repos some days, curious how grouping works when two sessions touch the same project on the same day, does it merge into one receipt or print one per session
Notchling
@galdayan One receipt for the day, one heading per project. Both sessions land under that project, each as its own line with its own start time, and the count next to the heading is how many lines are under it. Agent worktrees fold back into their parent repo rather than showing up as a separate project. In the week and month statements, identical lines across days merge into one with a count.
Love the receipt format – it turns an invisible day of agent work into something you can hold. What goes on it by default: tokens, tool calls, time? Have you seen people actually change how they work after seeing it?
Notchling
@kirillbush By default, only what you did: each session's title — or failing that, the commands it ran and the files it touched — grouped by project, with a start time. Deliberately no tokens or tool counts on the receipt. An earlier version had them and it read like a profiler, not like your day. Those numbers live on the dashboard instead.
On changing how people work: it's been a few days, so I won't claim that yet. The best signal so far is a reviewer saying they dread standups less.
FetchSandbox
The nm/otool proof is such a good call. Most privacy promises are just words, shipping the actual commands so anyone can verify it themselves is a completely different level of trust. 777 KB and zero network code is a real constraint, not a marketing line. Congrats on the launch, this is a clean idea.
Notchling
@rnagulapalle Appreciate that — the nm/otool thing felt important to actually show rather than just claim. Thanks for digging into it.
I counted my own ~/.claude/projects logs last week, and a token line per project is the thing I'd add. 5.86 billion tokens over 7 days, 98.88% of them cache reads.
One gotcha if you ever add it: a single streamed response writes several rows, so I had to dedupe by message id or it double counts.
Notchling
@abdullah_javaid3 5.86B tokens in a week is wild, that's a great idea for a metric line. I'll think about adding a token count per project — the dedupe-by-message-id gotcha is good to know, thanks for the heads up.