_repeat

Shared Repeat Protocol

Adds --repeat <N> --wait <seconds> and --tillend --wait <seconds> inter-session looping to any command. Uses ScheduleWakeup to pause between runs and .monomind/loops/<id>.json for dashboard tracking.

---

REPEAT PREAMBLE

Apply immediately after flag extraction, before any other command logic.

1. Extract repeat flags

Extract and remove these flags from $ARGUMENTS before passing the remainder to the command's own parser:

Flag	Variable	Default	Notes
--repeat <N>	repeat_count	0	N ≥ 2 activates repeat; N < 2 = disabled
--tillend	tillend_mode	false	Run until empty round (no findings, no actions). Overrides --repeat.
--maxruns <N>	tillend_maxruns	50	Safety cap for --tillend; stops after N runs even if AI hasn't signaled done
--wait <seconds>	wait_seconds	60	Minimum 60 (enforced by ScheduleWakeup)
--rep <N>	current_rep	absent	Internal; injected by ScheduleWakeup on continuation runs
--loop <id>	loop_id	absent	Internal; preserves loop identity across runs

If both --tillend and --repeat <N> are present, --tillend takes precedence.

2. If --rep N is present (continuation run)

• Set current_rep = N, loop_id from --loop <id>
• Set is_continuation = true
• The calling command MUST skip its empty-prompt check and intake when is_continuation = true

First — staleness guard (run before anything else):

Check whether the loop file exists and whether this wakeup is stale:

LOOP_FILE=".monomind/loops/${LOOP_ID}.json"
if [ ! -f "$LOOP_FILE" ]; then
  echo "[repeat] Stale wakeup: loop ${LOOP_ID} is already complete (state file gone). Skipping."
  # STOP — do not execute the command
else
  LOOP_CURRENT_REP=$(python3 -c "import json; print(json.load(open('${LOOP_FILE}'))['currentRep'])" 2>/dev/null \
    || jq -r '.currentRep // 0' "${LOOP_FILE}" 2>/dev/null || echo "0")
  if [ "${current_rep}" -lt "${LOOP_CURRENT_REP}" ]; then
    echo "[repeat] Stale wakeup: got --rep ${current_rep} but loop is already at rep ${LOOP_CURRENT_REP}. Skipping."
    # STOP — do not execute the command
  fi
fi