{
 "checker": "exact match (the harness's exact-match checker)",
 "curve": [
  {
   "id": "s01",
   "plain_usd_cum": {"high": 0.1241, "low": 0.0972, "mean": 0.1173, "n": 4},
   "recall_usd_cum_client": {"high": 0.0533, "low": 0.0525, "mean": 0.0529, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.0533, "low": 0.0525, "mean": 0.0529, "n": 4}
  },
  {
   "id": "s02",
   "plain_usd_cum": {"high": 0.1307, "low": 0.1055, "mean": 0.1242, "n": 4},
   "recall_usd_cum_client": {"high": 0.0605, "low": 0.0593, "mean": 0.0599, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.0605, "low": 0.0593, "mean": 0.0599, "n": 4}
  },
  {
   "id": "s03",
   "plain_usd_cum": {"high": 0.1372, "low": 0.1358, "mean": 0.1362, "n": 4},
   "recall_usd_cum_client": {"high": 0.0679, "low": 0.0671, "mean": 0.0674, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.0679, "low": 0.0671, "mean": 0.0674, "n": 4}
  },
  {
   "id": "r1",
   "plain_usd_cum": {"high": 0.9165, "low": 0.8916, "mean": 0.8985, "n": 4},
   "recall_usd_cum_client": {"high": 0.1813, "low": 0.1786, "mean": 0.1795, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.1813, "low": 0.1786, "mean": 0.1795, "n": 4}
  },
  {
   "id": "c1",
   "plain_usd_cum": {"high": 1.0636, "low": 1.0387, "mean": 1.0457, "n": 4},
   "recall_usd_cum_client": {"high": 0.2984, "low": 0.2949, "mean": 0.2961, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.2984, "low": 0.2949, "mean": 0.2961, "n": 4}
  },
  {
   "id": "r2",
   "plain_usd_cum": {"high": 2.0086, "low": 1.9455, "mean": 1.9808, "n": 4},
   "recall_usd_cum_client": {"high": 0.4357, "low": 0.4315, "mean": 0.434, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.4357, "low": 0.4315, "mean": 0.434, "n": 4}
  },
  {
   "id": "e1",
   "plain_usd_cum": {"high": 2.1766, "low": 2.1106, "mean": 2.148, "n": 4},
   "recall_usd_cum_client": {"high": 0.5086, "low": 0.4942, "mean": 0.5037, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.5086, "low": 0.4942, "mean": 0.5037, "n": 4}
  },
  {
   "id": "c2",
   "plain_usd_cum": {"high": 2.3712, "low": 2.3033, "mean": 2.3422, "n": 4},
   "recall_usd_cum_client": {"high": 0.6345, "low": 0.6203, "mean": 0.6296, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.6345, "low": 0.6203, "mean": 0.6296, "n": 4}
  },
  {
   "id": "r3",
   "plain_usd_cum": {"high": 3.2906, "low": 3.143, "mean": 3.2462, "n": 4},
   "recall_usd_cum_client": {"high": 0.8127, "low": 0.7721, "mean": 0.7877, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.8127, "low": 0.7721, "mean": 0.7877, "n": 4}
  },
  {
   "id": "e2",
   "plain_usd_cum": {"high": 3.5137, "low": 3.3621, "mean": 3.4672, "n": 4},
   "recall_usd_cum_client": {"high": 0.8827, "low": 0.8388, "mean": 0.8565, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.8827, "low": 0.8388, "mean": 0.8565, "n": 4}
  },
  {
   "id": "c3",
   "plain_usd_cum": {"high": 3.7515, "low": 3.5961, "mean": 3.7033, "n": 4},
   "recall_usd_cum_client": {"high": 1.0187, "low": 0.9745, "mean": 0.9921, "n": 4},
   "recall_usd_cum_ledger": {"high": 1.0187, "low": 0.9745, "mean": 0.9921, "n": 4}
  },
  {
   "id": "r4",
   "plain_usd_cum": {"high": 4.5438, "low": 4.3899, "mean": 4.4952, "n": 4},
   "recall_usd_cum_client": {"high": 1.2977, "low": 1.2516, "mean": 1.2705, "n": 4},
   "recall_usd_cum_ledger": {"high": 1.2977, "low": 1.2516, "mean": 1.2705, "n": 4}
  },
  {
   "id": "e3",
   "plain_usd_cum": {"high": 4.8195, "low": 4.6602, "mean": 4.7684, "n": 4},
   "recall_usd_cum_client": {"high": 1.8383, "low": 1.3398, "mean": 1.4738, "n": 4},
   "recall_usd_cum_ledger": {"high": 1.8383, "low": 1.3398, "mean": 1.4738, "n": 4}
  },
  {
   "id": "t1",
   "plain_usd_cum": {"high": 5.0997, "low": 4.9366, "mean": 5.047, "n": 4},
   "recall_usd_cum_client": {"high": 1.9871, "low": 1.9032, "mean": 1.9653, "n": 4},
   "recall_usd_cum_ledger": {"high": 1.9871, "low": 1.9032, "mean": 1.9653, "n": 4}
  },
  {
   "id": "d-delimiter",
   "plain_usd_cum": {"high": 5.2883, "low": 5.1228, "mean": 5.2333, "n": 4},
   "recall_usd_cum_client": {"high": 2.0336, "low": 1.9455, "mean": 2.0106, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.0336, "low": 1.9455, "mean": 2.0106, "n": 4}
  },
  {
   "id": "d-date",
   "plain_usd_cum": {"high": 5.4702, "low": 5.302, "mean": 5.4161, "n": 4},
   "recall_usd_cum_client": {"high": 2.0799, "low": 1.9861, "mean": 2.0548, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.0799, "low": 1.9861, "mean": 2.0548, "n": 4}
  },
  {
   "id": "t2",
   "plain_usd_cum": {"high": 5.6717, "low": 5.4933, "mean": 5.6159, "n": 4},
   "recall_usd_cum_client": {"high": 2.1585, "low": 2.0384, "mean": 2.1219, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.1585, "low": 2.0384, "mean": 2.1219, "n": 4}
  },
  {
   "id": "t3",
   "plain_usd_cum": {"high": 5.8578, "low": 5.673, "mean": 5.8006, "n": 4},
   "recall_usd_cum_client": {"high": 2.2009, "low": 2.0765, "mean": 2.1632, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.2009, "low": 2.0765, "mean": 2.1632, "n": 4}
  },
  {
   "id": "t4",
   "plain_usd_cum": {"high": 6.2214, "low": 6.0305, "mean": 6.1194, "n": 4},
   "recall_usd_cum_client": {"high": 2.2798, "low": 2.1168, "mean": 2.2326, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.2798, "low": 2.1168, "mean": 2.2326, "n": 4}
  },
  {
   "id": "t5",
   "plain_usd_cum": {"high": 6.4032, "low": 6.2083, "mean": 6.3011, "n": 4},
   "recall_usd_cum_client": {"high": 2.3222, "low": 2.1556, "mean": 2.2742, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.3222, "low": 2.1556, "mean": 2.2742, "n": 4}
  },
  {
   "id": "q-typed-codename",
   "plain_usd_cum": {"high": 6.4921, "low": 6.2953, "mean": 6.3907, "n": 4},
   "recall_usd_cum_client": {"high": 2.3418, "low": 2.1729, "mean": 2.2932, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.3418, "low": 2.1729, "mean": 2.2932, "n": 4}
  },
  {
   "id": "q-typed-pager",
   "plain_usd_cum": {"high": 6.5809, "low": 6.3823, "mean": 6.4803, "n": 4},
   "recall_usd_cum_client": {"high": 2.3614, "low": 2.1902, "mean": 2.3124, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.3614, "low": 2.1902, "mean": 2.3124, "n": 4}
  },
  {
   "id": "q-typed-port",
   "plain_usd_cum": {"high": 6.6697, "low": 6.4692, "mean": 6.5698, "n": 4},
   "recall_usd_cum_client": {"high": 2.3809, "low": 2.2075, "mean": 2.3314, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.3809, "low": 2.2075, "mean": 2.3314, "n": 4}
  },
  {
   "id": "q-typed-ticket",
   "plain_usd_cum": {"high": 6.7585, "low": 6.5561, "mean": 6.6594, "n": 4},
   "recall_usd_cum_client": {"high": 2.4005, "low": 2.2248, "mean": 2.3504, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.4005, "low": 2.2248, "mean": 2.3504, "n": 4}
  },
  {
   "id": "q-deleted-report-1",
   "plain_usd_cum": {"high": 6.8477, "low": 6.6434, "mean": 6.7494, "n": 4},
   "recall_usd_cum_client": {"high": 2.4247, "low": 2.2465, "mean": 2.374, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.4247, "low": 2.2465, "mean": 2.374, "n": 4}
  },
  {
   "id": "q-deleted-report-2",
   "plain_usd_cum": {"high": 6.9368, "low": 6.7306, "mean": 6.8393, "n": 4},
   "recall_usd_cum_client": {"high": 2.4456, "low": 2.268, "mean": 2.3951, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.4456, "low": 2.268, "mean": 2.3951, "n": 4}
  },
  {
   "id": "q-deleted-report-3",
   "plain_usd_cum": {"high": 7.026, "low": 6.8179, "mean": 6.9294, "n": 4},
   "recall_usd_cum_client": {"high": 2.4655, "low": 2.2865, "mean": 2.4148, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.4655, "low": 2.2865, "mean": 2.4148, "n": 4}
  },
  {
   "id": "q-const-series-max",
   "plain_usd_cum": {"high": 7.1152, "low": 6.9053, "mean": 7.0196, "n": 4},
   "recall_usd_cum_client": {"high": 2.4881, "low": 2.3063, "mean": 2.4365, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.4881, "low": 2.3063, "mean": 2.4365, "n": 4}
  },
  {
   "id": "q-const-activation-bytes",
   "plain_usd_cum": {"high": 7.2047, "low": 6.9929, "mean": 7.11, "n": 4},
   "recall_usd_cum_client": {"high": 2.515, "low": 2.3314, "mean": 2.4628, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.515, "low": 2.3314, "mean": 2.4628, "n": 4}
  },
  {
   "id": "q-const-activity-cap",
   "plain_usd_cum": {"high": 7.2942, "low": 7.0804, "mean": 7.2005, "n": 4},
   "recall_usd_cum_client": {"high": 2.5386, "low": 2.3526, "mean": 2.4854, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.5386, "low": 2.3526, "mean": 2.4854, "n": 4}
  }
 ],
 "legend": {"col.cr": "cache read tokens", "col.cw1h": "cache write tokens, 1 hour lifetime", "col.cw5": "cache write tokens, 5 minute lifetime", "col.in": "input tokens not read from or written to the cache", "col.missing": "a token column left out of a block is all zeros; model indexes into models", "col.out": "output tokens; null = not saved per call"},
 "mean_and_spread": {
  "compactions": {
   "plain": {"high": 0, "low": 0, "mean": 0, "n": 4},
   "recall": {"high": 0, "low": 0, "mean": 0, "n": 4}
  },
  "forwarded_tokens_max": {
   "plain": {"high": 441872, "low": 430901, "mean": 438707, "n": 4},
   "recall": {"high": 126989, "low": 126521, "mean": 126749.5, "n": 4}
  },
  "forwarded_tokens_mean": {
   "plain": {"high": 334320, "low": 327270, "mean": 330785.25, "n": 4},
   "recall": {"high": 78083, "low": 70468, "mean": 75590.75, "n": 4}
  },
  "probes_exact": {
   "plain": {"high": 10, "low": 10, "mean": 10, "n": 4},
   "recall": {"high": 6, "low": 5, "mean": 5.75, "n": 4}
  },
  "reasks": {
   "plain": {"high": 0, "low": 0, "mean": 0, "n": 4},
   "recall": {"high": 0, "low": 0, "mean": 0, "n": 4}
  },
  "recall_vs_plain_usd_pct": {"high": -64.4574, "low": -67.747, "mean": -65.47, "n": 4},
  "relayouts": {"high": 1, "low": 1, "mean": 1, "n": 4},
  "task_checks": {
   "plain": {"high": 8, "low": 8, "mean": 8, "n": 4},
   "recall": {"high": 8, "low": 8, "mean": 8, "n": 4}
  },
  "upstream_4xx": {
   "plain": {"high": 0, "low": 0, "mean": 0, "n": 4},
   "recall": {"high": 0, "low": 0, "mean": 0, "n": 4}
  },
  "usd": {
   "plain": {"high": 7.2942, "low": 7.0804, "mean": 7.2005, "n": 4},
   "recall": {"high": 2.5386, "low": 2.3526, "mean": 2.4854, "n": 4}
  }
 },
 "pairs": [
  {
   "deploy_after": {"at": "2026-10-03T04:19:05.720Z", "build_conclusion": "success", "build_id": "3ffa34b7-3e18-4a95-81c1-c525f25878da", "head": "98d7d704c8c498be7c384c43463163c5e17cdbbb", "worker_versions": "1884ba16-e6c4-45c6-8efb-940af6c65eaa"},
   "deploy_before": {"at": "2026-10-03T04:10:39.522Z", "build_conclusion": "success", "build_id": "3ffa34b7-3e18-4a95-81c1-c525f25878da", "head": "98d7d704c8c498be7c384c43463163c5e17cdbbb", "worker_versions": "1884ba16-e6c4-45c6-8efb-940af6c65eaa"},
   "deploy_changed": false,
   "deploy_checks": 60,
   "first": "plain",
   "pair": 1,
   "passes": false,
   "plain": {
    "cache_write_tokens": 437075,
    "client_reported_usd": 7.289,
    "compactions": 0,
    "curve": [
     {"id": "s01", "usd_client_cum": 0.124},
     {"id": "s02", "usd_client_cum": 0.1307},
     {"id": "s03", "usd_client_cum": 0.1372},
     {"id": "r1", "usd_client_cum": 0.8943},
     {"id": "c1", "usd_client_cum": 1.0418},
     {"id": "r2", "usd_client_cum": 1.986},
     {"id": "e1", "usd_client_cum": 2.1538},
     {"id": "c2", "usd_client_cum": 2.3485},
     {"id": "r3", "usd_client_cum": 3.2639},
     {"id": "e2", "usd_client_cum": 3.4847},
     {"id": "c3", "usd_client_cum": 3.7208},
     {"id": "r4", "usd_client_cum": 4.5115},
     {"id": "e3", "usd_client_cum": 4.7849},
     {"id": "t1", "usd_client_cum": 5.0638},
     {"id": "d-delimiter", "usd_client_cum": 5.2488},
     {"id": "d-date", "usd_client_cum": 5.4338},
     {"id": "t2", "usd_client_cum": 5.6365},
     {"id": "t3", "usd_client_cum": 5.822},
     {"id": "t4", "usd_client_cum": 6.1879},
     {"id": "t5", "usd_client_cum": 6.3716},
     {"id": "q-typed-codename", "usd_client_cum": 6.4628},
     {"id": "q-typed-pager", "usd_client_cum": 6.554},
     {"id": "q-typed-port", "usd_client_cum": 6.6453},
     {"id": "q-typed-ticket", "usd_client_cum": 6.7367},
     {"id": "q-deleted-report-1", "usd_client_cum": 6.8284},
     {"id": "q-deleted-report-2", "usd_client_cum": 6.9202},
     {"id": "q-deleted-report-3", "usd_client_cum": 7.0121},
     {"id": "q-const-series-max", "usd_client_cum": 7.1042},
     {"id": "q-const-activation-bytes", "usd_client_cum": 7.1966},
     {"id": "q-const-activity-cap", "usd_client_cum": 7.289}
    ],
    "decisions_asked": 2,
    "file": "pair-1-plain.json",
    "forwarded_tokens_max": 441717,
    "forwarded_tokens_mean": 332637,
    "partial": false,
    "probe_turns": {"n": 10, "tool_calls": 0, "usd": 0.9175, "write_tokens_max": 470},
    "probes": 10,
    "probes_by_kind": [
     {"exact": 4, "kind": "typed", "of": 4},
     {"exact": 3, "kind": "tool-output", "of": 3},
     {"exact": 3, "kind": "constant", "of": 3}
    ],
    "probes_contains": 10,
    "probes_of": 10,
    "reasks": 0,
    "recall_fail_rows": null,
    "recall_rows_by_phase": null,
    "recall_stats": null,
    "relayout_at": null,
    "relayouts": null,
    "seed": 424242,
    "steps_planned": 30,
    "steps_run": 30,
    "stopped": "",
    "task_checks": 8,
    "tenant": null,
    "tools_in_list": 25,
    "tools_list_constant": true,
    "upstream_4xx": 0,
    "usd": 7.289,
    "usd_client": 7.289,
    "usd_client_json": 7.289,
    "usd_ledger": null,
    "usd_transcript": 7.289,
    "void": false
   },
   "recall": {
    "cache_write_tokens": 188520,
    "client_reported_usd": 2.5386,
    "compactions": 0,
    "curve": [
     {"id": "s01", "usd_client_cum": 0.0526, "usd_ledger_cum": 0.0526},
     {"id": "s02", "usd_client_cum": 0.0595, "usd_ledger_cum": 0.0595},
     {"id": "s03", "usd_client_cum": 0.0671, "usd_ledger_cum": 0.0671},
     {"id": "r1", "usd_client_cum": 0.1786, "usd_ledger_cum": 0.1786},
     {"id": "c1", "usd_client_cum": 0.2949, "usd_ledger_cum": 0.2949},
     {"id": "r2", "usd_client_cum": 0.4331, "usd_ledger_cum": 0.4331},
     {"id": "e1", "usd_client_cum": 0.4942, "usd_ledger_cum": 0.4942},
     {"id": "c2", "usd_client_cum": 0.6203, "usd_ledger_cum": 0.6203},
     {"id": "r3", "usd_client_cum": 0.7721, "usd_ledger_cum": 0.7721},
     {"id": "e2", "usd_client_cum": 0.8388, "usd_ledger_cum": 0.8388},
     {"id": "c3", "usd_client_cum": 0.9745, "usd_ledger_cum": 0.9745},
     {"id": "r4", "usd_client_cum": 1.2516, "usd_ledger_cum": 1.2516},
     {"id": "e3", "usd_client_cum": 1.3398, "usd_ledger_cum": 1.3398},
     {"id": "t1", "usd_client_cum": 1.9871, "usd_ledger_cum": 1.9871},
     {"id": "d-delimiter", "usd_client_cum": 2.0336, "usd_ledger_cum": 2.0336},
     {"id": "d-date", "usd_client_cum": 2.0786, "usd_ledger_cum": 2.0786},
     {"id": "t2", "usd_client_cum": 2.1585, "usd_ledger_cum": 2.1585},
     {"id": "t3", "usd_client_cum": 2.2009, "usd_ledger_cum": 2.2009},
     {"id": "t4", "usd_client_cum": 2.2798, "usd_ledger_cum": 2.2798},
     {"id": "t5", "usd_client_cum": 2.3222, "usd_ledger_cum": 2.3222},
     {"id": "q-typed-codename", "usd_client_cum": 2.3418, "usd_ledger_cum": 2.3418},
     {"id": "q-typed-pager", "usd_client_cum": 2.3614, "usd_ledger_cum": 2.3614},
     {"id": "q-typed-port", "usd_client_cum": 2.3809, "usd_ledger_cum": 2.3809},
     {"id": "q-typed-ticket", "usd_client_cum": 2.4005, "usd_ledger_cum": 2.4005},
     {"id": "q-deleted-report-1", "usd_client_cum": 2.4247, "usd_ledger_cum": 2.4247},
     {"id": "q-deleted-report-2", "usd_client_cum": 2.4456, "usd_ledger_cum": 2.4456},
     {"id": "q-deleted-report-3", "usd_client_cum": 2.4655, "usd_ledger_cum": 2.4655},
     {"id": "q-const-series-max", "usd_client_cum": 2.4881, "usd_ledger_cum": 2.4881},
     {"id": "q-const-activation-bytes", "usd_client_cum": 2.515, "usd_ledger_cum": 2.515},
     {"id": "q-const-activity-cap", "usd_client_cum": 2.5386, "usd_ledger_cum": 2.5386}
    ],
    "decisions_asked": 2,
    "file": "pair-1-recall.json",
    "forwarded_tokens_max": 126989,
    "forwarded_tokens_mean": 78083,
    "partial": false,
    "probe_turns": {"n": 10, "tool_calls": 0, "usd": 0.2164, "write_tokens_max": 252},
    "probes": 6,
    "probes_by_kind": [
     {"exact": 4, "kind": "typed", "of": 4},
     {"exact": 2, "kind": "tool-output", "of": 3},
     {"exact": 0, "kind": "constant", "of": 3}
    ],
    "probes_contains": 7,
    "probes_of": 10,
    "reasks": 0,
    "recall_fail_rows": 0,
    "recall_rows_by_phase": {"recall": 58, "recall_C": 1, "recall_arch_a": 1, "recall_archived": 1, "recall_held": 27, "recall_read_ms": 58, "recall_read_src:store": 58, "recall_relayout": 1, "recall_trigger:first": 1, "recall_trigger:ski": 2},
    "recall_stats": {"archive.bytes": 1567212, "archive.episodes": 99, "reask": 1, "recall_hits": 0, "recall_judged": 0, "recalls": 0, "relayouts": 1, "saved_usd_recall": -0.353711, "triggers.first": 1, "triggers.ski": 2},
    "relayout_at": ["2026-10-03T04:14:49.815Z"],
    "relayouts": 1,
    "seed": 424242,
    "steps_planned": 30,
    "steps_run": 30,
    "stopped": "",
    "task_checks": 8,
    "tenant": "withheld (disposable test tenant)",
    "tools_in_list": 24,
    "tools_list_constant": true,
    "upstream_4xx": 0,
    "usd": 2.5386,
    "usd_client": 2.5386,
    "usd_client_json": 2.5386,
    "usd_ledger": 2.5386,
    "usd_transcript": 2.5386,
    "void": false
   },
   "recall_vs_plain_usd_pct": -65.1722,
   "seed": 424242,
   "valid": true,
   "why": ""
  },
  {
   "deploy_after": {"at": "2026-10-03T04:26:48.989Z", "build_conclusion": "success", "build_id": "3ffa34b7-3e18-4a95-81c1-c525f25878da", "head": "98d7d704c8c498be7c384c43463163c5e17cdbbb", "worker_versions": "1884ba16-e6c4-45c6-8efb-940af6c65eaa"},
   "deploy_before": {"at": "2026-10-03T04:19:09.202Z", "build_conclusion": "success", "build_id": "3ffa34b7-3e18-4a95-81c1-c525f25878da", "head": "98d7d704c8c498be7c384c43463163c5e17cdbbb", "worker_versions": "1884ba16-e6c4-45c6-8efb-940af6c65eaa"},
   "deploy_changed": false,
   "deploy_checks": 60,
   "first": "recall",
   "pair": 2,
   "passes": false,
   "plain": {
    "cache_write_tokens": 435696,
    "client_reported_usd": 7.2942,
    "compactions": 0,
    "curve": [
     {"id": "s01", "usd_client_cum": 0.1241},
     {"id": "s02", "usd_client_cum": 0.1304},
     {"id": "s03", "usd_client_cum": 0.1359},
     {"id": "r1", "usd_client_cum": 0.8916},
     {"id": "c1", "usd_client_cum": 1.0387},
     {"id": "r2", "usd_client_cum": 1.9831},
     {"id": "e1", "usd_client_cum": 2.151},
     {"id": "c2", "usd_client_cum": 2.3456},
     {"id": "r3", "usd_client_cum": 3.2906},
     {"id": "e2", "usd_client_cum": 3.5137},
     {"id": "c3", "usd_client_cum": 3.7515},
     {"id": "r4", "usd_client_cum": 4.5438},
     {"id": "e3", "usd_client_cum": 4.8195},
     {"id": "t1", "usd_client_cum": 5.0997},
     {"id": "d-delimiter", "usd_client_cum": 5.2883},
     {"id": "d-date", "usd_client_cum": 5.4702},
     {"id": "t2", "usd_client_cum": 5.6717},
     {"id": "t3", "usd_client_cum": 5.8578},
     {"id": "t4", "usd_client_cum": 6.2214},
     {"id": "t5", "usd_client_cum": 6.4032},
     {"id": "q-typed-codename", "usd_client_cum": 6.4921},
     {"id": "q-typed-pager", "usd_client_cum": 6.5809},
     {"id": "q-typed-port", "usd_client_cum": 6.6697},
     {"id": "q-typed-ticket", "usd_client_cum": 6.7585},
     {"id": "q-deleted-report-1", "usd_client_cum": 6.8477},
     {"id": "q-deleted-report-2", "usd_client_cum": 6.9368},
     {"id": "q-deleted-report-3", "usd_client_cum": 7.026},
     {"id": "q-const-series-max", "usd_client_cum": 7.1152},
     {"id": "q-const-activation-bytes", "usd_client_cum": 7.2047},
     {"id": "q-const-activity-cap", "usd_client_cum": 7.2942}
    ],
    "decisions_asked": 2,
    "file": "pair-2-plain.json",
    "forwarded_tokens_max": 440338,
    "forwarded_tokens_mean": 334320,
    "partial": false,
    "probe_turns": {"n": 10, "tool_calls": 0, "usd": 0.8911, "write_tokens_max": 126},
    "probes": 10,
    "probes_by_kind": [
     {"exact": 4, "kind": "typed", "of": 4},
     {"exact": 3, "kind": "tool-output", "of": 3},
     {"exact": 3, "kind": "constant", "of": 3}
    ],
    "probes_contains": 10,
    "probes_of": 10,
    "reasks": 0,
    "recall_fail_rows": null,
    "recall_rows_by_phase": null,
    "recall_stats": null,
    "relayout_at": null,
    "relayouts": null,
    "seed": 424242,
    "steps_planned": 30,
    "steps_run": 30,
    "stopped": "",
    "task_checks": 8,
    "tenant": null,
    "tools_in_list": 25,
    "tools_list_constant": true,
    "upstream_4xx": 0,
    "usd": 7.2942,
    "usd_client": 7.2942,
    "usd_client_json": 7.2942,
    "usd_ledger": null,
    "usd_transcript": 7.2942,
    "void": false
   },
   "recall": {
    "cache_write_tokens": 177214,
    "client_reported_usd": 2.3526,
    "compactions": 0,
    "curve": [
     {"id": "s01", "usd_client_cum": 0.0525, "usd_ledger_cum": 0.0525},
     {"id": "s02", "usd_client_cum": 0.0593, "usd_ledger_cum": 0.0593},
     {"id": "s03", "usd_client_cum": 0.0671, "usd_ledger_cum": 0.0671},
     {"id": "r1", "usd_client_cum": 0.1795, "usd_ledger_cum": 0.1795},
     {"id": "c1", "usd_client_cum": 0.2962, "usd_ledger_cum": 0.2962},
     {"id": "r2", "usd_client_cum": 0.4356, "usd_ledger_cum": 0.4356},
     {"id": "e1", "usd_client_cum": 0.5086, "usd_ledger_cum": 0.5086},
     {"id": "c2", "usd_client_cum": 0.6345, "usd_ledger_cum": 0.6345},
     {"id": "r3", "usd_client_cum": 0.8127, "usd_ledger_cum": 0.8127},
     {"id": "e2", "usd_client_cum": 0.8827, "usd_ledger_cum": 0.8827},
     {"id": "c3", "usd_client_cum": 1.0187, "usd_ledger_cum": 1.0187},
     {"id": "r4", "usd_client_cum": 1.2977, "usd_ledger_cum": 1.2977},
     {"id": "e3", "usd_client_cum": 1.8383, "usd_ledger_cum": 1.8383},
     {"id": "t1", "usd_client_cum": 1.9032, "usd_ledger_cum": 1.9032},
     {"id": "d-delimiter", "usd_client_cum": 1.9455, "usd_ledger_cum": 1.9455},
     {"id": "d-date", "usd_client_cum": 1.9861, "usd_ledger_cum": 1.9861},
     {"id": "t2", "usd_client_cum": 2.0384, "usd_ledger_cum": 2.0384},
     {"id": "t3", "usd_client_cum": 2.0765, "usd_ledger_cum": 2.0765},
     {"id": "t4", "usd_client_cum": 2.1168, "usd_ledger_cum": 2.1168},
     {"id": "t5", "usd_client_cum": 2.1556, "usd_ledger_cum": 2.1556},
     {"id": "q-typed-codename", "usd_client_cum": 2.1729, "usd_ledger_cum": 2.1729},
     {"id": "q-typed-pager", "usd_client_cum": 2.1902, "usd_ledger_cum": 2.1902},
     {"id": "q-typed-port", "usd_client_cum": 2.2075, "usd_ledger_cum": 2.2075},
     {"id": "q-typed-ticket", "usd_client_cum": 2.2248, "usd_ledger_cum": 2.2248},
     {"id": "q-deleted-report-1", "usd_client_cum": 2.2465, "usd_ledger_cum": 2.2465},
     {"id": "q-deleted-report-2", "usd_client_cum": 2.268, "usd_ledger_cum": 2.268},
     {"id": "q-deleted-report-3", "usd_client_cum": 2.2865, "usd_ledger_cum": 2.2865},
     {"id": "q-const-series-max", "usd_client_cum": 2.3063, "usd_ledger_cum": 2.3063},
     {"id": "q-const-activation-bytes", "usd_client_cum": 2.3314, "usd_ledger_cum": 2.3314},
     {"id": "q-const-activity-cap", "usd_client_cum": 2.3526, "usd_ledger_cum": 2.3526}
    ],
    "decisions_asked": 2,
    "file": "pair-2-recall.json",
    "forwarded_tokens_max": 126909,
    "forwarded_tokens_mean": 70468,
    "partial": false,
    "probe_turns": {"n": 10, "tool_calls": 0, "usd": 0.197, "write_tokens_max": 281},
    "probes": 5,
    "probes_by_kind": [
     {"exact": 4, "kind": "typed", "of": 4},
     {"exact": 1, "kind": "tool-output", "of": 3},
     {"exact": 0, "kind": "constant", "of": 3}
    ],
    "probes_contains": 6,
    "probes_of": 10,
    "reasks": 0,
    "recall_fail_rows": 0,
    "recall_rows_by_phase": {"recall": 56, "recall_C": 1, "recall_arch_a": 1, "recall_archived": 1, "recall_held": 25, "recall_read_ms": 56, "recall_read_src:store": 56, "recall_relayout": 1, "recall_trigger:first": 1, "recall_trigger:ski": 2},
    "recall_stats": {"archive.bytes": 1588495, "archive.episodes": 103, "reask": 1, "recall_hits": 0, "recall_judged": 0, "recalls": 0, "relayouts": 1, "saved_usd_recall": -0.217194, "triggers.first": 1, "triggers.ski": 2},
    "relayout_at": ["2026-10-03T04:22:43.805Z"],
    "relayouts": 1,
    "seed": 424242,
    "steps_planned": 30,
    "steps_run": 30,
    "stopped": "",
    "task_checks": 8,
    "tenant": "withheld (disposable test tenant)",
    "tools_in_list": 24,
    "tools_list_constant": true,
    "upstream_4xx": 0,
    "usd": 2.3526,
    "usd_client": 2.3526,
    "usd_client_json": 2.3526,
    "usd_ledger": 2.3526,
    "usd_transcript": 2.3526,
    "void": false
   },
   "recall_vs_plain_usd_pct": -67.747,
   "seed": 424242,
   "valid": true,
   "why": ""
  },
  {
   "deploy_after": {"at": "2026-10-03T04:34:22.739Z", "build_conclusion": "success", "build_id": "3ffa34b7-3e18-4a95-81c1-c525f25878da", "head": "98d7d704c8c498be7c384c43463163c5e17cdbbb", "worker_versions": "1884ba16-e6c4-45c6-8efb-940af6c65eaa"},
   "deploy_before": {"at": "2026-10-03T04:26:52.667Z", "build_conclusion": "success", "build_id": "3ffa34b7-3e18-4a95-81c1-c525f25878da", "head": "98d7d704c8c498be7c384c43463163c5e17cdbbb", "worker_versions": "1884ba16-e6c4-45c6-8efb-940af6c65eaa"},
   "deploy_changed": false,
   "deploy_checks": 60,
   "first": "plain",
   "pair": 3,
   "passes": false,
   "plain": {
    "cache_write_tokens": 426257,
    "client_reported_usd": 7.0804,
    "compactions": 0,
    "curve": [
     {"id": "s01", "usd_client_cum": 0.124},
     {"id": "s02", "usd_client_cum": 0.1304},
     {"id": "s03", "usd_client_cum": 0.1358},
     {"id": "r1", "usd_client_cum": 0.8917},
     {"id": "c1", "usd_client_cum": 1.0388},
     {"id": "r2", "usd_client_cum": 1.9455},
     {"id": "e1", "usd_client_cum": 2.1106},
     {"id": "c2", "usd_client_cum": 2.3033},
     {"id": "r3", "usd_client_cum": 3.143},
     {"id": "e2", "usd_client_cum": 3.3621},
     {"id": "c3", "usd_client_cum": 3.5961},
     {"id": "r4", "usd_client_cum": 4.3899},
     {"id": "e3", "usd_client_cum": 4.6602},
     {"id": "t1", "usd_client_cum": 4.9366},
     {"id": "d-delimiter", "usd_client_cum": 5.1228},
     {"id": "d-date", "usd_client_cum": 5.302},
     {"id": "t2", "usd_client_cum": 5.4933},
     {"id": "t3", "usd_client_cum": 5.673},
     {"id": "t4", "usd_client_cum": 6.0305},
     {"id": "t5", "usd_client_cum": 6.2083},
     {"id": "q-typed-codename", "usd_client_cum": 6.2953},
     {"id": "q-typed-pager", "usd_client_cum": 6.3823},
     {"id": "q-typed-port", "usd_client_cum": 6.4692},
     {"id": "q-typed-ticket", "usd_client_cum": 6.5561},
     {"id": "q-deleted-report-1", "usd_client_cum": 6.6434},
     {"id": "q-deleted-report-2", "usd_client_cum": 6.7306},
     {"id": "q-deleted-report-3", "usd_client_cum": 6.8179},
     {"id": "q-const-series-max", "usd_client_cum": 6.9053},
     {"id": "q-const-activation-bytes", "usd_client_cum": 6.9929},
     {"id": "q-const-activity-cap", "usd_client_cum": 7.0804}
    ],
    "decisions_asked": 2,
    "file": "pair-3-plain.json",
    "forwarded_tokens_max": 430901,
    "forwarded_tokens_mean": 327270,
    "partial": false,
    "probe_turns": {"n": 10, "tool_calls": 0, "usd": 0.8723, "write_tokens_max": 126},
    "probes": 10,
    "probes_by_kind": [
     {"exact": 4, "kind": "typed", "of": 4},
     {"exact": 3, "kind": "tool-output", "of": 3},
     {"exact": 3, "kind": "constant", "of": 3}
    ],
    "probes_contains": 10,
    "probes_of": 10,
    "reasks": 0,
    "recall_fail_rows": null,
    "recall_rows_by_phase": null,
    "recall_stats": null,
    "relayout_at": null,
    "relayouts": null,
    "seed": 424242,
    "steps_planned": 30,
    "steps_run": 30,
    "stopped": "",
    "task_checks": 8,
    "tenant": null,
    "tools_in_list": 25,
    "tools_list_constant": true,
    "upstream_4xx": 0,
    "usd": 7.0804,
    "usd_client": 7.0804,
    "usd_client_json": 7.0804,
    "usd_ledger": null,
    "usd_transcript": 7.0804,
    "void": false
   },
   "recall": {
    "cache_write_tokens": 188440,
    "client_reported_usd": 2.5133,
    "compactions": 0,
    "curve": [
     {"id": "s01", "usd_client_cum": 0.0533, "usd_ledger_cum": 0.0533},
     {"id": "s02", "usd_client_cum": 0.0605, "usd_ledger_cum": 0.0605},
     {"id": "s03", "usd_client_cum": 0.0676, "usd_ledger_cum": 0.0676},
     {"id": "r1", "usd_client_cum": 0.1788, "usd_ledger_cum": 0.1788},
     {"id": "c1", "usd_client_cum": 0.295, "usd_ledger_cum": 0.295},
     {"id": "r2", "usd_client_cum": 0.4315, "usd_ledger_cum": 0.4315},
     {"id": "e1", "usd_client_cum": 0.5036, "usd_ledger_cum": 0.5036},
     {"id": "c2", "usd_client_cum": 0.6294, "usd_ledger_cum": 0.6294},
     {"id": "r3", "usd_client_cum": 0.7813, "usd_ledger_cum": 0.7813},
     {"id": "e2", "usd_client_cum": 0.8504, "usd_ledger_cum": 0.8504},
     {"id": "c3", "usd_client_cum": 0.9858, "usd_ledger_cum": 0.9858},
     {"id": "r4", "usd_client_cum": 1.2659, "usd_ledger_cum": 1.2659},
     {"id": "e3", "usd_client_cum": 1.3588, "usd_ledger_cum": 1.3588},
     {"id": "t1", "usd_client_cum": 1.987, "usd_ledger_cum": 1.987},
     {"id": "d-delimiter", "usd_client_cum": 2.0333, "usd_ledger_cum": 2.0333},
     {"id": "d-date", "usd_client_cum": 2.0799, "usd_ledger_cum": 2.0799},
     {"id": "t2", "usd_client_cum": 2.1359, "usd_ledger_cum": 2.1359},
     {"id": "t3", "usd_client_cum": 2.178, "usd_ledger_cum": 2.178},
     {"id": "t4", "usd_client_cum": 2.257, "usd_ledger_cum": 2.257},
     {"id": "t5", "usd_client_cum": 2.2995, "usd_ledger_cum": 2.2995},
     {"id": "q-typed-codename", "usd_client_cum": 2.3191, "usd_ledger_cum": 2.3191},
     {"id": "q-typed-pager", "usd_client_cum": 2.3388, "usd_ledger_cum": 2.3388},
     {"id": "q-typed-port", "usd_client_cum": 2.3583, "usd_ledger_cum": 2.3583},
     {"id": "q-typed-ticket", "usd_client_cum": 2.3779, "usd_ledger_cum": 2.3779},
     {"id": "q-deleted-report-1", "usd_client_cum": 2.4022, "usd_ledger_cum": 2.4022},
     {"id": "q-deleted-report-2", "usd_client_cum": 2.4229, "usd_ledger_cum": 2.4229},
     {"id": "q-deleted-report-3", "usd_client_cum": 2.4429, "usd_ledger_cum": 2.4429},
     {"id": "q-const-series-max", "usd_client_cum": 2.4649, "usd_ledger_cum": 2.4649},
     {"id": "q-const-activation-bytes", "usd_client_cum": 2.4909, "usd_ledger_cum": 2.4909},
     {"id": "q-const-activity-cap", "usd_client_cum": 2.5133, "usd_ledger_cum": 2.5133}
    ],
    "decisions_asked": 2,
    "file": "pair-3-recall.json",
    "forwarded_tokens_max": 126521,
    "forwarded_tokens_mean": 76719,
    "partial": false,
    "probe_turns": {"n": 10, "tool_calls": 0, "usd": 0.2136, "write_tokens_max": 229},
    "probes": 6,
    "probes_by_kind": [
     {"exact": 4, "kind": "typed", "of": 4},
     {"exact": 2, "kind": "tool-output", "of": 3},
     {"exact": 0, "kind": "constant", "of": 3}
    ],
    "probes_contains": 7,
    "probes_of": 10,
    "reasks": 0,
    "recall_fail_rows": 0,
    "recall_rows_by_phase": {"recall": 57, "recall_C": 1, "recall_arch_a": 1, "recall_archived": 1, "recall_held": 26, "recall_read_ms": 57, "recall_read_src:store": 57, "recall_relayout": 1, "recall_trigger:first": 1, "recall_trigger:ski": 2},
    "recall_stats": {"archive.bytes": 1564036, "archive.episodes": 98, "reask": 1, "recall_hits": 0, "recall_judged": 0, "recalls": 0, "relayouts": 1, "saved_usd_recall": -0.368532, "triggers.first": 1, "triggers.ski": 2},
    "relayout_at": ["2026-10-03T04:30:32.719Z"],
    "relayouts": 1,
    "seed": 424242,
    "steps_planned": 30,
    "steps_run": 30,
    "stopped": "",
    "task_checks": 8,
    "tenant": "withheld (disposable test tenant)",
    "tools_in_list": 24,
    "tools_list_constant": true,
    "upstream_4xx": 0,
    "usd": 2.5133,
    "usd_client": 2.5133,
    "usd_client_json": 2.5133,
    "usd_ledger": 2.5133,
    "usd_transcript": 2.5133,
    "void": false
   },
   "recall_vs_plain_usd_pct": -64.5034,
   "seed": 424242,
   "valid": true,
   "why": ""
  },
  {
   "deploy_after": {"at": "2026-10-03T04:41:59.599Z", "build_conclusion": "success", "build_id": "3ffa34b7-3e18-4a95-81c1-c525f25878da", "head": "98d7d704c8c498be7c384c43463163c5e17cdbbb", "worker_versions": "1884ba16-e6c4-45c6-8efb-940af6c65eaa"},
   "deploy_before": {"at": "2026-10-03T04:34:26.361Z", "build_conclusion": "success", "build_id": "3ffa34b7-3e18-4a95-81c1-c525f25878da", "head": "98d7d704c8c498be7c384c43463163c5e17cdbbb", "worker_versions": "1884ba16-e6c4-45c6-8efb-940af6c65eaa"},
   "deploy_changed": false,
   "deploy_checks": 60,
   "first": "recall",
   "pair": 4,
   "passes": false,
   "plain": {
    "cache_write_tokens": 440318,
    "client_reported_usd": 7.1382,
    "compactions": 0,
    "curve": [
     {"id": "s01", "usd_client_cum": 0.0972},
     {"id": "s02", "usd_client_cum": 0.1055},
     {"id": "s03", "usd_client_cum": 0.1359},
     {"id": "r1", "usd_client_cum": 0.9165},
     {"id": "c1", "usd_client_cum": 1.0636},
     {"id": "r2", "usd_client_cum": 2.0086},
     {"id": "e1", "usd_client_cum": 2.1766},
     {"id": "c2", "usd_client_cum": 2.3712},
     {"id": "r3", "usd_client_cum": 3.2872},
     {"id": "e2", "usd_client_cum": 3.5083},
     {"id": "c3", "usd_client_cum": 3.7446},
     {"id": "r4", "usd_client_cum": 4.5355},
     {"id": "e3", "usd_client_cum": 4.8092},
     {"id": "t1", "usd_client_cum": 5.0879},
     {"id": "d-delimiter", "usd_client_cum": 5.2733},
     {"id": "d-date", "usd_client_cum": 5.4582},
     {"id": "t2", "usd_client_cum": 5.6622},
     {"id": "t3", "usd_client_cum": 5.8496},
     {"id": "t4", "usd_client_cum": 6.0378},
     {"id": "t5", "usd_client_cum": 6.2214},
     {"id": "q-typed-codename", "usd_client_cum": 6.3126},
     {"id": "q-typed-pager", "usd_client_cum": 6.4038},
     {"id": "q-typed-port", "usd_client_cum": 6.495},
     {"id": "q-typed-ticket", "usd_client_cum": 6.5863},
     {"id": "q-deleted-report-1", "usd_client_cum": 6.678},
     {"id": "q-deleted-report-2", "usd_client_cum": 6.7697},
     {"id": "q-deleted-report-3", "usd_client_cum": 6.8616},
     {"id": "q-const-series-max", "usd_client_cum": 6.9536},
     {"id": "q-const-activation-bytes", "usd_client_cum": 7.0459},
     {"id": "q-const-activity-cap", "usd_client_cum": 7.1382}
    ],
    "decisions_asked": 2,
    "file": "pair-4-plain.json",
    "forwarded_tokens_max": 441872,
    "forwarded_tokens_mean": 328914,
    "partial": false,
    "probe_turns": {"n": 10, "tool_calls": 0, "usd": 0.9168, "write_tokens_max": 457},
    "probes": 10,
    "probes_by_kind": [
     {"exact": 4, "kind": "typed", "of": 4},
     {"exact": 3, "kind": "tool-output", "of": 3},
     {"exact": 3, "kind": "constant", "of": 3}
    ],
    "probes_contains": 10,
    "probes_of": 10,
    "reasks": 0,
    "recall_fail_rows": null,
    "recall_rows_by_phase": null,
    "recall_stats": null,
    "relayout_at": null,
    "relayouts": null,
    "seed": 424242,
    "steps_planned": 30,
    "steps_run": 30,
    "stopped": "",
    "task_checks": 8,
    "tenant": null,
    "tools_in_list": 25,
    "tools_list_constant": true,
    "upstream_4xx": 0,
    "usd": 7.1382,
    "usd_client": 7.1382,
    "usd_client_json": 7.1382,
    "usd_ledger": null,
    "usd_transcript": 7.1382,
    "void": false
   },
   "recall": {
    "cache_write_tokens": 188780,
    "client_reported_usd": 2.5371,
    "compactions": 0,
    "curve": [
     {"id": "s01", "usd_client_cum": 0.0532, "usd_ledger_cum": 0.0532},
     {"id": "s02", "usd_client_cum": 0.0603, "usd_ledger_cum": 0.0603},
     {"id": "s03", "usd_client_cum": 0.0679, "usd_ledger_cum": 0.0679},
     {"id": "r1", "usd_client_cum": 0.1813, "usd_ledger_cum": 0.1813},
     {"id": "c1", "usd_client_cum": 0.2984, "usd_ledger_cum": 0.2984},
     {"id": "r2", "usd_client_cum": 0.4357, "usd_ledger_cum": 0.4357},
     {"id": "e1", "usd_client_cum": 0.5082, "usd_ledger_cum": 0.5082},
     {"id": "c2", "usd_client_cum": 0.6341, "usd_ledger_cum": 0.6341},
     {"id": "r3", "usd_client_cum": 0.7846, "usd_ledger_cum": 0.7846},
     {"id": "e2", "usd_client_cum": 0.8539, "usd_ledger_cum": 0.8539},
     {"id": "c3", "usd_client_cum": 0.9894, "usd_ledger_cum": 0.9894},
     {"id": "r4", "usd_client_cum": 1.2667, "usd_ledger_cum": 1.2667},
     {"id": "e3", "usd_client_cum": 1.3584, "usd_ledger_cum": 1.3584},
     {"id": "t1", "usd_client_cum": 1.9838, "usd_ledger_cum": 1.9838},
     {"id": "d-delimiter", "usd_client_cum": 2.03, "usd_ledger_cum": 2.03},
     {"id": "d-date", "usd_client_cum": 2.0744, "usd_ledger_cum": 2.0744},
     {"id": "t2", "usd_client_cum": 2.1547, "usd_ledger_cum": 2.1547},
     {"id": "t3", "usd_client_cum": 2.1972, "usd_ledger_cum": 2.1972},
     {"id": "t4", "usd_client_cum": 2.2767, "usd_ledger_cum": 2.2767},
     {"id": "t5", "usd_client_cum": 2.3194, "usd_ledger_cum": 2.3194},
     {"id": "q-typed-codename", "usd_client_cum": 2.3391, "usd_ledger_cum": 2.3391},
     {"id": "q-typed-pager", "usd_client_cum": 2.3591, "usd_ledger_cum": 2.3591},
     {"id": "q-typed-port", "usd_client_cum": 2.3787, "usd_ledger_cum": 2.3787},
     {"id": "q-typed-ticket", "usd_client_cum": 2.3984, "usd_ledger_cum": 2.3984},
     {"id": "q-deleted-report-1", "usd_client_cum": 2.4225, "usd_ledger_cum": 2.4225},
     {"id": "q-deleted-report-2", "usd_client_cum": 2.4438, "usd_ledger_cum": 2.4438},
     {"id": "q-deleted-report-3", "usd_client_cum": 2.4644, "usd_ledger_cum": 2.4644},
     {"id": "q-const-series-max", "usd_client_cum": 2.4867, "usd_ledger_cum": 2.4867},
     {"id": "q-const-activation-bytes", "usd_client_cum": 2.514, "usd_ledger_cum": 2.514},
     {"id": "q-const-activity-cap", "usd_client_cum": 2.5371, "usd_ledger_cum": 2.5371}
    ],
    "decisions_asked": 2,
    "file": "pair-4-recall.json",
    "forwarded_tokens_max": 126579,
    "forwarded_tokens_mean": 77093,
    "partial": false,
    "probe_turns": {"n": 10, "tool_calls": 0, "usd": 0.2178, "write_tokens_max": 221},
    "probes": 6,
    "probes_by_kind": [
     {"exact": 4, "kind": "typed", "of": 4},
     {"exact": 2, "kind": "tool-output", "of": 3},
     {"exact": 0, "kind": "constant", "of": 3}
    ],
    "probes_contains": 7,
    "probes_of": 10,
    "reasks": 0,
    "recall_fail_rows": 0,
    "recall_rows_by_phase": {"recall": 58, "recall_C": 1, "recall_arch_a": 1, "recall_archived": 1, "recall_held": 27, "recall_read_ms": 58, "recall_read_src:store": 58, "recall_relayout": 1, "recall_trigger:first": 1, "recall_trigger:ski": 2},
    "recall_stats": {"archive.bytes": 1566216, "archive.episodes": 100, "reask": 1, "recall_hits": 0, "recall_judged": 0, "recalls": 0, "relayouts": 1, "saved_usd_recall": -0.360159, "triggers.first": 1, "triggers.ski": 2},
    "relayout_at": ["2026-10-03T04:38:03.427Z"],
    "relayouts": 1,
    "seed": 424242,
    "steps_planned": 30,
    "steps_run": 30,
    "stopped": "",
    "task_checks": 8,
    "tenant": "withheld (disposable test tenant)",
    "tools_in_list": 24,
    "tools_list_constant": true,
    "upstream_4xx": 0,
    "usd": 2.5371,
    "usd_client": 2.5371,
    "usd_client_json": 2.5371,
    "usd_ledger": 2.5371,
    "usd_transcript": 2.5371,
    "void": false
   },
   "recall_vs_plain_usd_pct": -64.4574,
   "seed": 424242,
   "valid": true,
   "why": ""
  }
 ],
 "pairs_passing": 0,
 "pairs_planned": 4,
 "pairs_run": 4,
 "pairs_valid": 4,
 "prereg": "pre-registered in gateway commit f2b5c567 (2026-10-03T07:01:17+03:00), before the first pair",
 "prereg_check": {"commit": "f2b5c567", "committed": "2026-10-03T04:01:17.000Z", "plan": "prereg/long-session-prereg.md", "run_started": "2026-10-03T04:11:12.962Z", "run_started_from": "the earliest 'at' of the arm files these pairs name"},
 "schema": "context-mode.benchmark-evidence/2",
 "ship": "FAIL",
 "ship_rule": "each pair: recall exact probes >= plain exact probes AND recall $ < plain $ AND task checks equal AND recall upstream 4xx = 0; ship = all 4 pairs present, valid and passing",
 "spend": {
  "runs": {"dry-plain": 0.8929, "dry-recall": 0.2031, "dry-recall-b4k": 0.4044, "dry2-recall": 0.1724, "dry2-recall-b4k": 0.2732, "dry3-plain": 0.9047, "p1-plain": 7.289, "p1-recall": 2.5386, "p2-plain": 7.2942, "p2-recall": 2.3526, "p3-plain": 7.0804, "p3-recall": 2.5133, "p4-plain": 7.1382, "p4-recall": 2.5371}
 }
}