Your AI coding day,
in one report.
DayLedger reads your Claude Code sessions and turns each day into a clean, scored work report — sessions, output, focus, and cost — then files every one into a browsable Library. It all runs on your machine. Your transcripts never leave it.
macOS & Linux · Python 3.8+ · reads ~/.claude · only dependency is PyYAML
A standup you didn't have to write.
Most days blur together. DayLedger gives each one a record — what you shipped, where your effort went, and what to pick up next — generated from the work itself.
Every day becomes one report — sessions, output, focus, and cost distilled into a single DayLedger Score with day-over-day trends.
Reports are filed into a Library you can browse — filter by day, type, or project, with a score-over-time chart.
Auto-generates a wrap-up at 10pm and a morning brief at 8am. You wake up to yesterday, already written.
It reads your Claude logs straight off disk. Nothing is uploaded. The only dependency is PyYAML.
A self-contained app that opens offline — plus an always-on local viewer at localhost:7099.
Insights, focuses, and an assignment ledger pulled from the work itself — not a generic activity counter.
Up and running in a minute.
No account, no cloud, no telemetry. Three steps and you're reading your first report.
Grab the free bundle and unzip it. One small Python package, no account, no signup.
./install.sh sets up a private workspace, generates your first report, and adds a launcher to your Desktop.
Browse every report in one place. Add --schedule to auto-generate them at 10pm and 8am.
unzip dayledger.zip && cd dayledger
./install.sh # set up + your first report
./install.sh --schedule # …and auto-generate at 10pm & 8am (macOS)Private by design.
DayLedger runs entirely on your computer. It reads your Claude logs locally and writes reports to a folder you own — no servers, no accounts, no analytics. Open-and-read the code if you like; the whole engine is a small, dependency-light Python package.