Changelog

[Unreleased]

Show local-day read and write totals with their recorded timezone, labelled as totals so far for the current day.
Open the dashboard on a live three-hour written-volume graph with a read/write toggle and selected-point inspection.
Browse history by day from the keyboard with [, ], g date entry, and t for today and the live view.

Collect every three minutes on systemd and runit, plot live points at their actual timestamps, and retain three-minute detail for 14 days before durable summaries.
Withhold the endurance outlook until one full local observation day has usable evidence, then show it with categorical confidence.

Keep midnight-spanning activity once as shared boundary evidence instead of adding it to both days.
Publish derived hour and day history consistently with each collection run.

Make usage-history axes, units, UTC boundaries, active 7/14/30/90-day window, gaps, partial periods, stacked write attribution, and hourly drill-down explicit; keep live graph data refreshed on the current three-minute cadence.

Use sentence case throughout the dashboard and add persistent keyboard and sudo guidance.
Share read-only status acquisition between the CLI and TUI, preserving unknown monitoring state and store-fault recovery guidance.
Use one authenticated action path for the CLI and TUI; let valid collection runs finish without the former 30-second dashboard cutoff.
Remove the unused sparkline and habit-bar rendering path while retaining the interactive history graph.
Align all package formats on the MIT license and Python 3.10 minimum; include the license text in native packages.

Correct observation-store directory permissions in native packages, including repair of older runit installations during upgrade.

Show trustworthy first-use history, qualifying-day progress, and confidence in the dashboard.
Add an interactive daily-writes graph with hourly drill-down and evidence-anchored scenario windows.
Repair retained observation history safely and keep its retention state visible.
Present a unified monitoring status, drive-health context, and Fenris identity in the dashboard.
Remember accessible colour and motion preferences and keep the dashboard usable at constrained terminal sizes.

Add Fenris identity and a polkit authentication notice to the dashboard.
Clarify monitoring continuity, deliberate pauses, and quitting in the dashboard and status output.
Add per-release notes with installation, verification, and rollback guidance.