calendar.snapshot.biweekly (#805)

Parent: calendar.access.verify Link: seed.show/calendar.snapshot.biweekly Views: 0

Content

Build a calendar snapshot covering last week and this week (14-day window centered on today). Query all accessible calendars from the cached calendar list, merging results into a single chronological timeline.

For each event include: start time, end time, title, calendar name (if multiple calendars), location (if set), and whether it's a recurring instance. Mark all-day events separately from timed events. For recurring events, expand instances — don't just show the recurrence rule. Convert all times to the user's local timezone, noting the timezone used.

Structure the output as a day-by-day timeline: group events under date headers, sorted chronologically within each day. For days with no events, include the date with "no events" rather than skipping it — gaps in the timeline matter for understanding workload patterns.

Flag notable patterns: days with 5+ events, back-to-back meetings with no gaps, events outside normal hours (before 8am or after 7pm), and any scheduling conflicts (overlapping timed events on the same calendar). If an event has a video conference link (Google Meet, Zoom, Teams), include the join URL.

Handle provider-specific quirks: Google Calendar returns times in RFC3339 with timezone offset; Microsoft Graph returns dateTime + timeZone as separate fields that must be combined; CalDAV returns DTSTART/DTEND in iCalendar format which may be floating (no timezone), UTC, or with TZID. Declined events should be excluded unless explicitly requested. Cancelled events must be filtered out — Google marks these with status:cancelled, Graph uses isCancelled:true, CalDAV uses STATUS:CANCELLED.

If a calendar returns more than 200 events in the window (possible with heavily shared calendars), paginate and warn the user about the volume rather than silently truncating.

Conjugation

Evals — 8/8 ✅