fleet

Fleet endpoint and state (raw)

!PORT=${FLEETDECK_PORT:-4711}; case "$PORT" in ''|*[!0-9]*|??????*) PORT=4711;; *) if [ "$PORT" -lt 1 ] || [ "$PORT" -gt 65535 ]; then PORT=4711; fi;; esac; printf 'FLEET_BOARD_URL=http://127.0.0.1:%s/\n' "$PORT"; TOKEN=$(cat "${FLEETDECK_HOME:-$HOME/.fleetdeck}/token" 2>/dev/null); curl -sf -m 2 ${TOKEN:+-H "Authorization: Bearer $TOKEN"} "http://127.0.0.1:$PORT/state" || echo FLEET_DAEMON_DOWN

Your task

Report on the fleet using ONLY the endpoint and raw state above. The first line is the exact board URL for this environment; use its value verbatim. Print exactly this, nothing more:

If the raw state is FLEET_DAEMON_DOWN (or empty / not JSON):

Board: fleet daemon not running — it starts with your next session

Otherwise a compact summary (aim for under 15 lines):

1. Board: <the FLEET_BOARD_URL value above>
2. Sessions grouped by column, in this order: needsyou, working, verifying, queued, idle, offline. One line per column that has sessions: working (2): NOVA — fleetdeck (main) · editing derive.mjs, ... using each session's callsign, repo_name, branch, and note/task when present. Skip empty columns; summarize offline as a count only (e.g. offline: 3).
3. Conflicts: if conflicts is non-empty, one line per conflict: ⚠ <rel_path> — <severity> — sessions: <callsigns or ids>. Otherwise conflicts: none.
4. Mail: mail_pending is a per-session count map {<session_id>: <count>} (sessions with no queued mail appear with 0). Compute N as the total across all sessions — the sum of every value in the map. If N > 0: mail: N queued — followed by a per-route breakdown derived from mail_meta (only list routes that have queued mail): watcher = wakes immediately, pane = typed into the pane after a grace window, turn-boundary = delivered at the next turn boundary, offline-queued = held until the session is resumed. Example: mail: 3 queued — 1 wakes immediately (watcher), 1 typed into pane, 1 held until resume (offline). Otherwise omit the line.