From letterdog
Prepare for meetings from Letterdog calendar events — the user's personal self-hosted mail/calendar/contacts (JMAP), not Google Calendar. Use when the user asks what a meeting is about, what to read, who is attending, what links or attachments matter, or for a concise prep brief.
How this skill is triggered — by the user, by Claude, or both
Slash command
/letterdog:letterdog-calendar-meeting-prepThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Tool mechanics, confirmation flow, and CLI routing live in the `letterdog-calendar` skill. This
Tool mechanics, confirmation flow, and CLI routing live in the letterdog-calendar skill. This
skill covers only the prep workflow.
If the user gives an event id, call read_events with that id (base or synthetic instance id;
projection: "full" when links/description/participants matter). Otherwise search_events over a
bounded window (after/before LocalDateTime + time_zone; expand: true when the meeting is a
recurring occurrence), matching the user's words via title or text, then read_events on the
candidates. When multiple events match, disambiguate by time, title, organizer, and calendar before
doing deeper work.
Use the full event object as the primary source: title, description, start/end, timezone, location,
virtual-location links, participants and their statuses, organizer, recurrence, status. Resolve
unfamiliar attendees with search_people / read_contacts. Related mail is often the real agenda:
search_emails for the organizer or subject keywords, read_thread for the latest state (mail
tools are documented in letterdog-mail). Raw iCalendar fidelity, if it ever matters, routes to the
CLI (letterdog dav get). Search adjacent calendar context only when useful: nearby events with the
same participants, repeated titles, project names.
Keep the output practical:
Do not fabricate agenda details. Label inferences as inferences.
npx claudepluginhub solastrius/letterdog --plugin letterdogCreates structured, bite-sized implementation plans from specs or requirements before writing code. Useful for breaking down multi-step tasks into testable steps with file structure and task boundaries.