nsimonfr/event-departure-planning

Event Departure Planning

Use this workflow for prompts such as: “pour mon prochain événement, à quelle heure dois-je partir ?” Read the personal calendar live, identify the next relevant event and its location, calculate a realistic departure plan, then state the answer concisely.

Default transport rule

• Default to bicycle.
• If the event is explicitly with Alfie, default to metro / public transport.
• If the attendee or transport mode is unclear and that choice materially changes the plan, ask one concise question before estimating.
• If the event has no usable location, ask for it. Do not invent one.

Workflow

1. Get current time with date and query the personal Nextcloud calendar live using caldav-calendar.

   • For “next event”, inspect today and the following days until a timed event with a location is found.
   • Use the current residence: 52 boulevard Arago, 75013 Paris through the end of October 2026; otherwise 82 rue Alexandre Dumas, Paris.
   • Completion: event name, start time, location, and origin are known.

2. Get a route estimate.

   • Bicycle: use ~/.hermes/skills/productivity/maps/scripts/maps_client.py distance ORIGIN --to DESTINATION --mode cycling.
   • Metro/public transport: use a current route source (prefer official RATP/Île-de-France Mobilités; web search is acceptable if official routing is unavailable). Capture walking legs, expected wait/connection time, and final walk.
   • The local Maps skill does not provide public-transit routing; never label a car/walk result as metro.
   • Completion: report the route duration and its components/assumptions.

3. Build the departure time.

   • Include 5 minutes to get downstairs / leave the building unless the user says otherwise.
   • Add a practical arrival margin by default: 15 minutes for ordinary appointments; 20 minutes for cinema, station/airport, first-time locations, or events where punctuality matters.
   • Add a modest transport contingency: at least 5 minutes for cycling; for metro, include a realistic wait/connection allowance and add 5–10 minutes beyond the nominal route when live disruption data is unavailable.
   • Calculate event start − (downstairs + route + contingency + arrival margin) using Python; do not mental-calculate.
   • Completion: a clock time is produced and is before the event.

4. Answer in this compact format:

   **[Event] — [start time]**
   Mode: [bike / metro]
   Trajet estimé : ~[N min] ([brief components]).
   Marge incluse : [downstairs + contingency + arrival buffer].
   **Pars vers [HH:MM].**
   

   Mention that estimates without live traffic/disruption data may vary. Do not create or modify calendar events.

Common Pitfalls

• Do not reuse a stale calendar result: read the calendar in the current turn.
• Do not assume a permanent home address; apply the date rule.
• Do not treat OSRM cycling as live traffic or transit data.
• Do not silently choose metro when the user did not mention Alfie; bicycle is the default.
• Do not omit the travel duration or the included margin.

Verification Checklist

• [ ] Calendar was read live.
• [ ] Event location and start time are explicit.
• [ ] Correct origin for the date was used.
• [ ] Mode follows the bicycle/Alfie/clarification rule.
• [ ] Departure calculation includes 5 minutes downstairs and a buffer.
• [ ] Reply includes both route time and departure time.