{
 "checker": "exact match (the harness's exact-match checker)",
 "curve": [
  {
   "id": "s01",
   "plain_usd_cum": {"high": 0.1245, "low": 0.0973, "mean": 0.1175, "n": 4},
   "recall_usd_cum_client": {"high": 0.0538, "low": 0.0358, "mean": 0.0487, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.0538, "low": 0.0358, "mean": 0.0487, "n": 4}
  },
  {
   "id": "s02",
   "plain_usd_cum": {"high": 0.1315, "low": 0.1057, "mean": 0.1245, "n": 4},
   "recall_usd_cum_client": {"high": 0.0611, "low": 0.0446, "mean": 0.0564, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.0611, "low": 0.0446, "mean": 0.0564, "n": 4}
  },
  {
   "id": "s03",
   "plain_usd_cum": {"high": 0.1373, "low": 0.1358, "mean": 0.1363, "n": 4},
   "recall_usd_cum_client": {"high": 0.0675, "low": 0.0671, "mean": 0.0672, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.0675, "low": 0.0671, "mean": 0.0672, "n": 4}
  },
  {
   "id": "r1",
   "plain_usd_cum": {"high": 0.9174, "low": 0.8916, "mean": 0.8986, "n": 4},
   "recall_usd_cum_client": {"high": 0.1808, "low": 0.1767, "mean": 0.1785, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.1808, "low": 0.1767, "mean": 0.1785, "n": 4}
  },
  {
   "id": "c1",
   "plain_usd_cum": {"high": 1.0646, "low": 1.0386, "mean": 1.0458, "n": 4},
   "recall_usd_cum_client": {"high": 0.2978, "low": 0.2923, "mean": 0.2945, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.2978, "low": 0.2923, "mean": 0.2945, "n": 4}
  },
  {
   "id": "r2",
   "plain_usd_cum": {"high": 2.0098, "low": 1.981, "mean": 1.9893, "n": 4},
   "recall_usd_cum_client": {"high": 0.439, "low": 0.4294, "mean": 0.4332, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.439, "low": 0.4294, "mean": 0.4332, "n": 4}
  },
  {
   "id": "e1",
   "plain_usd_cum": {"high": 2.178, "low": 2.1509, "mean": 2.1588, "n": 4},
   "recall_usd_cum_client": {"high": 0.5022, "low": 0.4909, "mean": 0.4955, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.5022, "low": 0.4909, "mean": 0.4955, "n": 4}
  },
  {
   "id": "c2",
   "plain_usd_cum": {"high": 2.3728, "low": 2.3454, "mean": 2.3535, "n": 4},
   "recall_usd_cum_client": {"high": 0.6284, "low": 0.6159, "mean": 0.6213, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.6284, "low": 0.6159, "mean": 0.6213, "n": 4}
  },
  {
   "id": "r3",
   "plain_usd_cum": {"high": 3.3201, "low": 3.2904, "mean": 3.2999, "n": 4},
   "recall_usd_cum_client": {"high": 0.8693, "low": 0.7837, "mean": 0.8299, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.8693, "low": 0.7837, "mean": 0.8299, "n": 4}
  },
  {
   "id": "e2",
   "plain_usd_cum": {"high": 3.5436, "low": 3.5138, "mean": 3.524, "n": 4},
   "recall_usd_cum_client": {"high": 0.9396, "low": 0.855, "mean": 0.9003, "n": 4},
   "recall_usd_cum_ledger": {"high": 0.9396, "low": 0.855, "mean": 0.9003, "n": 4}
  },
  {
   "id": "c3",
   "plain_usd_cum": {"high": 3.7815, "low": 3.7515, "mean": 3.7619, "n": 4},
   "recall_usd_cum_client": {"high": 1.0759, "low": 0.9908, "mean": 1.0361, "n": 4},
   "recall_usd_cum_ledger": {"high": 1.0759, "low": 0.9908, "mean": 1.0361, "n": 4}
  },
  {
   "id": "r4",
   "plain_usd_cum": {"high": 4.5744, "low": 4.5437, "mean": 4.5544, "n": 4},
   "recall_usd_cum_client": {"high": 1.3539, "low": 1.2698, "mean": 1.3227, "n": 4},
   "recall_usd_cum_ledger": {"high": 1.3539, "low": 1.2698, "mean": 1.3227, "n": 4}
  },
  {
   "id": "e3",
   "plain_usd_cum": {"high": 4.8504, "low": 4.8195, "mean": 4.8309, "n": 4},
   "recall_usd_cum_client": {"high": 1.9563, "low": 1.3625, "mean": 1.784, "n": 4},
   "recall_usd_cum_ledger": {"high": 1.9563, "low": 1.3625, "mean": 1.784, "n": 4}
  },
  {
   "id": "t1",
   "plain_usd_cum": {"high": 5.1295, "low": 5.0997, "mean": 5.1107, "n": 4},
   "recall_usd_cum_client": {"high": 2.0487, "low": 1.9767, "mean": 2.0081, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.0487, "low": 1.9767, "mean": 2.0081, "n": 4}
  },
  {
   "id": "d-delimiter",
   "plain_usd_cum": {"high": 5.3187, "low": 5.2894, "mean": 5.2985, "n": 4},
   "recall_usd_cum_client": {"high": 2.0972, "low": 2.0212, "mean": 2.0605, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.0972, "low": 2.0212, "mean": 2.0605, "n": 4}
  },
  {
   "id": "d-date",
   "plain_usd_cum": {"high": 5.5005, "low": 5.4712, "mean": 5.4821, "n": 4},
   "recall_usd_cum_client": {"high": 2.1434, "low": 2.0633, "mean": 2.1058, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.1434, "low": 2.0633, "mean": 2.1058, "n": 4}
  },
  {
   "id": "t2",
   "plain_usd_cum": {"high": 5.6988, "low": 5.6723, "mean": 5.683, "n": 4},
   "recall_usd_cum_client": {"high": 2.2301, "low": 2.1163, "mean": 2.1713, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.2301, "low": 2.1163, "mean": 2.1713, "n": 4}
  },
  {
   "id": "t3",
   "plain_usd_cum": {"high": 5.8849, "low": 5.8557, "mean": 5.8686, "n": 4},
   "recall_usd_cum_client": {"high": 2.2797, "low": 2.1539, "mean": 2.2131, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.2797, "low": 2.1539, "mean": 2.2131, "n": 4}
  },
  {
   "id": "t4",
   "plain_usd_cum": {"high": 6.2366, "low": 6.0439, "mean": 6.1468, "n": 4},
   "recall_usd_cum_client": {"high": 2.3309, "low": 2.1944, "mean": 2.2574, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.3309, "low": 2.1944, "mean": 2.2574, "n": 4}
  },
  {
   "id": "t5",
   "plain_usd_cum": {"high": 6.4216, "low": 6.2256, "mean": 6.3299, "n": 4},
   "recall_usd_cum_client": {"high": 2.3839, "low": 2.2339, "mean": 2.3006, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.3839, "low": 2.2339, "mean": 2.3006, "n": 4}
  },
  {
   "id": "q-typed-codename",
   "plain_usd_cum": {"high": 6.5135, "low": 6.3145, "mean": 6.4201, "n": 4},
   "recall_usd_cum_client": {"high": 2.4113, "low": 2.2526, "mean": 2.3216, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.4113, "low": 2.2526, "mean": 2.3216, "n": 4}
  },
  {
   "id": "q-typed-pager",
   "plain_usd_cum": {"high": 6.6054, "low": 6.4033, "mean": 6.5104, "n": 4},
   "recall_usd_cum_client": {"high": 2.4373, "low": 2.2705, "mean": 2.3418, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.4373, "low": 2.2705, "mean": 2.3418, "n": 4}
  },
  {
   "id": "q-typed-port",
   "plain_usd_cum": {"high": 6.6973, "low": 6.4921, "mean": 6.6006, "n": 4},
   "recall_usd_cum_client": {"high": 2.4633, "low": 2.2883, "mean": 2.362, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.4633, "low": 2.2883, "mean": 2.362, "n": 4}
  },
  {
   "id": "q-typed-ticket",
   "plain_usd_cum": {"high": 6.7893, "low": 6.581, "mean": 6.691, "n": 4},
   "recall_usd_cum_client": {"high": 2.4895, "low": 2.3062, "mean": 2.3823, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.4895, "low": 2.3062, "mean": 2.3823, "n": 4}
  },
  {
   "id": "q-deleted-report-1",
   "plain_usd_cum": {"high": 6.8817, "low": 6.6701, "mean": 6.7816, "n": 4},
   "recall_usd_cum_client": {"high": 2.5698, "low": 2.3365, "mean": 2.4462, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.592, "low": 2.355, "mean": 2.4661, "n": 4}
  },
  {
   "id": "q-deleted-report-2",
   "plain_usd_cum": {"high": 6.9741, "low": 6.7592, "mean": 6.8723, "n": 4},
   "recall_usd_cum_client": {"high": 2.6519, "low": 2.4618, "mean": 2.5489, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.6991, "low": 2.5025, "mean": 2.5918, "n": 4}
  },
  {
   "id": "q-deleted-report-3",
   "plain_usd_cum": {"high": 7.0667, "low": 6.8484, "mean": 6.963, "n": 4},
   "recall_usd_cum_client": {"high": 2.7355, "low": 2.4894, "mean": 2.5872, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.8073, "low": 2.5535, "mean": 2.642, "n": 4}
  },
  {
   "id": "q-const-series-max",
   "plain_usd_cum": {"high": 7.1594, "low": 6.9376, "mean": 7.0538, "n": 4},
   "recall_usd_cum_client": {"high": 2.7822, "low": 2.5358, "mean": 2.6241, "n": 4},
   "recall_usd_cum_ledger": {"high": 2.9118, "low": 2.6201, "mean": 2.7256, "n": 4}
  },
  {
   "id": "q-const-activation-bytes",
   "plain_usd_cum": {"high": 7.2524, "low": 7.0271, "mean": 7.1449, "n": 4},
   "recall_usd_cum_client": {"high": 2.8379, "low": 2.5629, "mean": 2.6601, "n": 4},
   "recall_usd_cum_ledger": {"high": 3.0347, "low": 2.67, "mean": 2.7932, "n": 4}
  },
  {
   "id": "q-const-activity-cap",
   "plain_usd_cum": {"high": 7.3454, "low": 7.1166, "mean": 7.2361, "n": 4},
   "recall_usd_cum_client": {"high": 2.9024, "low": 2.6984, "mean": 2.7595, "n": 4},
   "recall_usd_cum_ledger": {"high": 3.1708, "low": 2.8395, "mean": 2.9667, "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": 445217, "low": 440319, "mean": 442650, "n": 4},
   "recall": {"high": 128152, "low": 127159, "mean": 127822.25, "n": 4}
  },
  "forwarded_tokens_mean": {
   "plain": {"high": 335236, "low": 330461, "mean": 332849.75, "n": 4},
   "recall": {"high": 84165, "low": 73663, "mean": 77911.25, "n": 4}
  },
  "probes_exact": {
   "plain": {"high": 10, "low": 10, "mean": 10, "n": 4},
   "recall": {"high": 10, "low": 8, "mean": 9.5, "n": 4}
  },
  "probes_found_recorded": {"high": 4, "low": 1, "mean": 2.5, "n": 4},
  "probes_with_search": {"high": 6, "low": 5, "mean": 5.5, "n": 4},
  "reasks": {
   "plain": {"high": 0, "low": 0, "mean": 0, "n": 4},
   "recall": {"high": 0, "low": 0, "mean": 0, "n": 4}
  },
  "recall_net_usd": {"high": -0.1767, "low": -0.3994, "mean": -0.2883, "n": 4},
  "recall_vs_plain_usd_pct": {"high": -55.6314, "low": -61.292, "mean": -58.9815, "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": 7.3454, "low": 7.1166, "mean": 7.2361, "n": 4},
   "recall": {"high": 3.1708, "low": 2.8395, "mean": 2.9667, "n": 4}
  }
 },
 "pairs": [
  {
   "deploy_after": {"at": "2026-10-03T12:28:28.644Z", "build_conclusion": "success", "build_id": "17e928e5-71cb-46aa-9592-f006f61b7588", "head": "466ebe20b9dcd717c1e6cf2459a85bcbf200d810", "worker_versions": "e19f36ae-4553-437e-9b87-b983d242976d"},
   "deploy_before": {"at": "2026-10-03T12:19:52.314Z", "build_conclusion": "success", "build_id": "17e928e5-71cb-46aa-9592-f006f61b7588", "head": "466ebe20b9dcd717c1e6cf2459a85bcbf200d810", "worker_versions": "e19f36ae-4553-437e-9b87-b983d242976d"},
   "deploy_changed": false,
   "deploy_checks": 60,
   "first": "plain",
   "pair": 1,
   "passes": true,
   "plain": {
    "cache_write_tokens": 438864,
    "client_reported_usd": 7.1465,
    "compactions": 0,
    "curve": [
     {"id": "s01", "usd_client_cum": 0.0973},
     {"id": "s02", "usd_client_cum": 0.1057},
     {"id": "s03", "usd_client_cum": 0.1361},
     {"id": "r1", "usd_client_cum": 0.9174},
     {"id": "c1", "usd_client_cum": 1.0646},
     {"id": "r2", "usd_client_cum": 2.0098},
     {"id": "e1", "usd_client_cum": 2.178},
     {"id": "c2", "usd_client_cum": 2.3728},
     {"id": "r3", "usd_client_cum": 3.3201},
     {"id": "e2", "usd_client_cum": 3.5436},
     {"id": "c3", "usd_client_cum": 3.7815},
     {"id": "r4", "usd_client_cum": 4.5744},
     {"id": "e3", "usd_client_cum": 4.8504},
     {"id": "t1", "usd_client_cum": 5.1295},
     {"id": "d-delimiter", "usd_client_cum": 5.3187},
     {"id": "d-date", "usd_client_cum": 5.5005},
     {"id": "t2", "usd_client_cum": 5.6988},
     {"id": "t3", "usd_client_cum": 5.8849},
     {"id": "t4", "usd_client_cum": 6.0738},
     {"id": "t5", "usd_client_cum": 6.2554},
     {"id": "q-typed-codename", "usd_client_cum": 6.3442},
     {"id": "q-typed-pager", "usd_client_cum": 6.4331},
     {"id": "q-typed-port", "usd_client_cum": 6.5219},
     {"id": "q-typed-ticket", "usd_client_cum": 6.6108},
     {"id": "q-deleted-report-1", "usd_client_cum": 6.6999},
     {"id": "q-deleted-report-2", "usd_client_cum": 6.7891},
     {"id": "q-deleted-report-3", "usd_client_cum": 6.8782},
     {"id": "q-const-series-max", "usd_client_cum": 6.9675},
     {"id": "q-const-activation-bytes", "usd_client_cum": 7.057},
     {"id": "q-const-activity-cap", "usd_client_cum": 7.1465}
    ],
    "decisions_asked": 2,
    "file": "pair-1-plain.json",
    "forwarded_tokens_max": 440411,
    "forwarded_tokens_mean": 330559,
    "partial": false,
    "probe_turns": {"n": 10, "tool_calls": 0, "usd": 0.8913, "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_net_usd": null,
    "recall_rows_by_phase": null,
    "recall_stats": null,
    "relayout_at": null,
    "relayouts": null,
    "search": 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": 7.1465,
    "usd_client": 7.1465,
    "usd_client_json": 7.1465,
    "usd_ledger": null,
    "usd_transcript": 7.1465,
    "void": false
   },
   "recall": {
    "cache_write_tokens": 201061,
    "client_reported_usd": 2.9024,
    "compactions": 0,
    "curve": [
     {"id": "s01", "usd_client_cum": 0.0538, "usd_ledger_cum": 0.0538},
     {"id": "s02", "usd_client_cum": 0.0611, "usd_ledger_cum": 0.0611},
     {"id": "s03", "usd_client_cum": 0.0675, "usd_ledger_cum": 0.0675},
     {"id": "r1", "usd_client_cum": 0.1808, "usd_ledger_cum": 0.1808},
     {"id": "c1", "usd_client_cum": 0.2978, "usd_ledger_cum": 0.2978},
     {"id": "r2", "usd_client_cum": 0.439, "usd_ledger_cum": 0.439},
     {"id": "e1", "usd_client_cum": 0.5022, "usd_ledger_cum": 0.5022},
     {"id": "c2", "usd_client_cum": 0.6284, "usd_ledger_cum": 0.6284},
     {"id": "r3", "usd_client_cum": 0.7837, "usd_ledger_cum": 0.7837},
     {"id": "e2", "usd_client_cum": 0.855, "usd_ledger_cum": 0.855},
     {"id": "c3", "usd_client_cum": 0.9908, "usd_ledger_cum": 0.9908},
     {"id": "r4", "usd_client_cum": 1.2698, "usd_ledger_cum": 1.2698},
     {"id": "e3", "usd_client_cum": 1.3625, "usd_ledger_cum": 1.3625},
     {"id": "t1", "usd_client_cum": 2.0287, "usd_ledger_cum": 2.0287},
     {"id": "d-delimiter", "usd_client_cum": 2.0833, "usd_ledger_cum": 2.0833},
     {"id": "d-date", "usd_client_cum": 2.1321, "usd_ledger_cum": 2.1321},
     {"id": "t2", "usd_client_cum": 2.2301, "usd_ledger_cum": 2.2301},
     {"id": "t3", "usd_client_cum": 2.2797, "usd_ledger_cum": 2.2797},
     {"id": "t4", "usd_client_cum": 2.3309, "usd_ledger_cum": 2.3309},
     {"id": "t5", "usd_client_cum": 2.3839, "usd_ledger_cum": 2.3839},
     {"id": "q-typed-codename", "usd_client_cum": 2.4113, "usd_ledger_cum": 2.4113},
     {"id": "q-typed-pager", "usd_client_cum": 2.4373, "usd_ledger_cum": 2.4373},
     {"id": "q-typed-port", "usd_client_cum": 2.4633, "usd_ledger_cum": 2.4633},
     {"id": "q-typed-ticket", "usd_client_cum": 2.4895, "usd_ledger_cum": 2.4895},
     {"id": "q-deleted-report-1", "usd_client_cum": 2.5698, "usd_ledger_cum": 2.592},
     {"id": "q-deleted-report-2", "usd_client_cum": 2.6519, "usd_ledger_cum": 2.6991},
     {"id": "q-deleted-report-3", "usd_client_cum": 2.7355, "usd_ledger_cum": 2.8073},
     {"id": "q-const-series-max", "usd_client_cum": 2.7822, "usd_ledger_cum": 2.9118},
     {"id": "q-const-activation-bytes", "usd_client_cum": 2.8379, "usd_ledger_cum": 3.0347},
     {"id": "q-const-activity-cap", "usd_client_cum": 2.9024, "usd_ledger_cum": 3.1708}
    ],
    "decisions_asked": 2,
    "file": "pair-1-recall.json",
    "forwarded_tokens_max": 127834,
    "forwarded_tokens_mean": 84165,
    "partial": false,
    "probe_turns": {"n": 10, "tool_calls": 0, "usd": 0.5183, "write_tokens_max": 880},
    "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": 0,
    "recall_net_usd": -0.3994,
    "recall_rows_by_phase": {"recall": 56, "recall_C": 1, "recall_age_ms": 3, "recall_arch_a": 1, "recall_archived": 1, "recall_held": 25, "recall_n": 3, "recall_read_ms": 56, "recall_read_src:store": 56, "recall_relayout": 1, "recall_trigger:first": 1, "recall_trigger:ski": 2},
    "recall_stats": {"archive.bytes": 1568180, "archive.episodes": 98, "median_recall_age_ms": 389002, "reask": 0, "recall_hits": 0, "recall_judged": 0, "recalls": 27, "relayouts": 1, "saved_usd_recall": -0.399376, "triggers.first": 1, "triggers.ski": 2},
    "relayout_at": ["2026-10-03T12:23:48.409Z"],
    "relayouts": 1,
    "search": {
     "by_probe": [
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "typed", "probe": "typed-codename", "query_calls": 0},
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "typed", "probe": "typed-pager", "query_calls": 0},
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "typed", "probe": "typed-port", "query_calls": 0},
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "typed", "probe": "typed-ticket", "query_calls": 0},
      {"calls": 1, "client_saw_search": false, "docid_calls": 1, "exact": true, "found_recorded": true, "found_replay": false, "kind": "tool-output", "probe": "deleted-report-1", "query_calls": 0},
      {"calls": 1, "client_saw_search": false, "docid_calls": 1, "exact": true, "found_recorded": false, "found_replay": false, "kind": "tool-output", "probe": "deleted-report-2", "query_calls": 0},
      {"calls": 1, "client_saw_search": false, "docid_calls": 1, "exact": true, "found_recorded": false, "found_replay": false, "kind": "tool-output", "probe": "deleted-report-3", "query_calls": 0},
      {"calls": 2, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": true, "kind": "constant", "probe": "const-series-max", "query_calls": 2},
      {"calls": 3, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": true, "kind": "constant", "probe": "const-activation-bytes", "query_calls": 3},
      {"calls": 4, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": true, "kind": "constant", "probe": "const-activity-cap", "query_calls": 4}
     ],
     "exact_with_search": 6,
     "exact_without_search": 4,
     "execs_in_probes": 12,
     "execs_total": 12,
     "probes_found_recorded": 1,
     "probes_found_replay": 3,
     "probes_with_search": 6
    },
    "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": 3.1708,
    "usd_client": 2.9024,
    "usd_client_json": 2.9024,
    "usd_ledger": 3.1708,
    "usd_transcript": 2.9024,
    "void": false
   },
   "recall_vs_plain_usd_pct": -55.6314,
   "seed": 424242,
   "valid": true,
   "why": ""
  },
  {
   "deploy_after": {"at": "2026-10-03T12:38:37.382Z", "build_conclusion": "success", "build_id": "17e928e5-71cb-46aa-9592-f006f61b7588", "head": "466ebe20b9dcd717c1e6cf2459a85bcbf200d810", "worker_versions": "e19f36ae-4553-437e-9b87-b983d242976d"},
   "deploy_before": {"at": "2026-10-03T12:29:04.497Z", "build_conclusion": "success", "build_id": "17e928e5-71cb-46aa-9592-f006f61b7588", "head": "466ebe20b9dcd717c1e6cf2459a85bcbf200d810", "worker_versions": "e19f36ae-4553-437e-9b87-b983d242976d"},
   "deploy_changed": false,
   "deploy_checks": 60,
   "first": "recall",
   "pair": 2,
   "passes": true,
   "plain": {
    "cache_write_tokens": 440576,
    "client_reported_usd": 7.3454,
    "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": 0.8917},
     {"id": "c1", "usd_client_cum": 1.0388},
     {"id": "r2", "usd_client_cum": 1.9835},
     {"id": "e1", "usd_client_cum": 2.1536},
     {"id": "c2", "usd_client_cum": 2.3483},
     {"id": "r3", "usd_client_cum": 3.2949},
     {"id": "e2", "usd_client_cum": 3.5191},
     {"id": "c3", "usd_client_cum": 3.757},
     {"id": "r4", "usd_client_cum": 4.5493},
     {"id": "e3", "usd_client_cum": 4.8256},
     {"id": "t1", "usd_client_cum": 5.1061},
     {"id": "d-delimiter", "usd_client_cum": 5.2921},
     {"id": "d-date", "usd_client_cum": 5.4785},
     {"id": "t2", "usd_client_cum": 5.6829},
     {"id": "t3", "usd_client_cum": 5.8694},
     {"id": "t4", "usd_client_cum": 6.2366},
     {"id": "t5", "usd_client_cum": 6.4216},
     {"id": "q-typed-codename", "usd_client_cum": 6.5135},
     {"id": "q-typed-pager", "usd_client_cum": 6.6054},
     {"id": "q-typed-port", "usd_client_cum": 6.6973},
     {"id": "q-typed-ticket", "usd_client_cum": 6.7893},
     {"id": "q-deleted-report-1", "usd_client_cum": 6.8817},
     {"id": "q-deleted-report-2", "usd_client_cum": 6.9741},
     {"id": "q-deleted-report-3", "usd_client_cum": 7.0667},
     {"id": "q-const-series-max", "usd_client_cum": 7.1594},
     {"id": "q-const-activation-bytes", "usd_client_cum": 7.2524},
     {"id": "q-const-activity-cap", "usd_client_cum": 7.3454}
    ],
    "decisions_asked": 2,
    "file": "pair-2-plain.json",
    "forwarded_tokens_max": 445217,
    "forwarded_tokens_mean": 335143,
    "partial": false,
    "probe_turns": {"n": 10, "tool_calls": 0, "usd": 0.9238, "write_tokens_max": 461},
    "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_net_usd": null,
    "recall_rows_by_phase": null,
    "recall_stats": null,
    "relayout_at": null,
    "relayouts": null,
    "search": 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": 7.3454,
    "usd_client": 7.3454,
    "usd_client_json": 7.3454,
    "usd_ledger": null,
    "usd_transcript": 7.3454,
    "void": false
   },
   "recall": {
    "cache_write_tokens": 179978,
    "client_reported_usd": 2.6984,
    "compactions": 0,
    "curve": [
     {"id": "s01", "usd_client_cum": 0.0529, "usd_ledger_cum": 0.0529},
     {"id": "s02", "usd_client_cum": 0.0598, "usd_ledger_cum": 0.0598},
     {"id": "s03", "usd_client_cum": 0.0671, "usd_ledger_cum": 0.0671},
     {"id": "r1", "usd_client_cum": 0.1767, "usd_ledger_cum": 0.1767},
     {"id": "c1", "usd_client_cum": 0.2923, "usd_ledger_cum": 0.2923},
     {"id": "r2", "usd_client_cum": 0.4307, "usd_ledger_cum": 0.4307},
     {"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": 0.8693, "usd_ledger_cum": 0.8693},
     {"id": "e2", "usd_client_cum": 0.9396, "usd_ledger_cum": 0.9396},
     {"id": "c3", "usd_client_cum": 1.0759, "usd_ledger_cum": 1.0759},
     {"id": "r4", "usd_client_cum": 1.3539, "usd_ledger_cum": 1.3539},
     {"id": "e3", "usd_client_cum": 1.9097, "usd_ledger_cum": 1.9097},
     {"id": "t1", "usd_client_cum": 1.9767, "usd_ledger_cum": 1.9767},
     {"id": "d-delimiter", "usd_client_cum": 2.0212, "usd_ledger_cum": 2.0212},
     {"id": "d-date", "usd_client_cum": 2.0633, "usd_ledger_cum": 2.0633},
     {"id": "t2", "usd_client_cum": 2.1163, "usd_ledger_cum": 2.1163},
     {"id": "t3", "usd_client_cum": 2.1539, "usd_ledger_cum": 2.1539},
     {"id": "t4", "usd_client_cum": 2.1944, "usd_ledger_cum": 2.1944},
     {"id": "t5", "usd_client_cum": 2.2339, "usd_ledger_cum": 2.2339},
     {"id": "q-typed-codename", "usd_client_cum": 2.2526, "usd_ledger_cum": 2.2526},
     {"id": "q-typed-pager", "usd_client_cum": 2.2705, "usd_ledger_cum": 2.2705},
     {"id": "q-typed-port", "usd_client_cum": 2.2883, "usd_ledger_cum": 2.2883},
     {"id": "q-typed-ticket", "usd_client_cum": 2.3062, "usd_ledger_cum": 2.3062},
     {"id": "q-deleted-report-1", "usd_client_cum": 2.3365, "usd_ledger_cum": 2.355},
     {"id": "q-deleted-report-2", "usd_client_cum": 2.4618, "usd_ledger_cum": 2.5025},
     {"id": "q-deleted-report-3", "usd_client_cum": 2.4894, "usd_ledger_cum": 2.5535},
     {"id": "q-const-series-max", "usd_client_cum": 2.5358, "usd_ledger_cum": 2.689},
     {"id": "q-const-activation-bytes", "usd_client_cum": 2.5629, "usd_ledger_cum": 2.7352},
     {"id": "q-const-activity-cap", "usd_client_cum": 2.6984, "usd_ledger_cum": 2.9579}
    ],
    "decisions_asked": 2,
    "file": "pair-2-recall.json",
    "forwarded_tokens_max": 128144,
    "forwarded_tokens_mean": 73944,
    "partial": false,
    "probe_turns": {"n": 10, "tool_calls": 0, "usd": 0.4644, "write_tokens_max": 155},
    "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": 0,
    "recall_net_usd": -0.2059,
    "recall_rows_by_phase": {"recall": 59, "recall_C": 1, "recall_age_ms": 4, "recall_arch_a": 1, "recall_archived": 1, "recall_held": 27, "recall_n": 4, "recall_read_ms": 59, "recall_read_src:store": 59, "recall_relayout": 1, "recall_trigger:first": 1, "recall_trigger:ski": 2},
    "recall_stats": {"archive.bytes": 1560569, "archive.episodes": 98, "median_recall_age_ms": 433276, "reask": 1, "recall_hits": 0, "recall_judged": 0, "recalls": 28, "relayouts": 1, "saved_usd_recall": -0.205884, "triggers.first": 1, "triggers.ski": 2},
    "relayout_at": ["2026-10-03T12:33:09.543Z"],
    "relayouts": 1,
    "search": {
     "by_probe": [
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "typed", "probe": "typed-codename", "query_calls": 0},
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "typed", "probe": "typed-pager", "query_calls": 0},
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "typed", "probe": "typed-port", "query_calls": 0},
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "typed", "probe": "typed-ticket", "query_calls": 0},
      {"calls": 1, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": true, "kind": "tool-output", "probe": "deleted-report-1", "query_calls": 1},
      {"calls": 2, "client_saw_search": false, "docid_calls": 2, "exact": true, "found_recorded": true, "found_replay": false, "kind": "tool-output", "probe": "deleted-report-2", "query_calls": 0},
      {"calls": 1, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": true, "kind": "tool-output", "probe": "deleted-report-3", "query_calls": 1},
      {"calls": 3, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": true, "kind": "constant", "probe": "const-series-max", "query_calls": 3},
      {"calls": 1, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": true, "found_replay": true, "kind": "constant", "probe": "const-activation-bytes", "query_calls": 1},
      {"calls": 3, "client_saw_search": false, "docid_calls": 2, "exact": true, "found_recorded": true, "found_replay": true, "kind": "constant", "probe": "const-activity-cap", "query_calls": 1}
     ],
     "exact_with_search": 6,
     "exact_without_search": 4,
     "execs_in_probes": 11,
     "execs_total": 11,
     "probes_found_recorded": 3,
     "probes_found_replay": 5,
     "probes_with_search": 6
    },
    "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": 2.9579,
    "usd_client": 2.6984,
    "usd_client_json": 2.6984,
    "usd_ledger": 2.9579,
    "usd_transcript": 2.6984,
    "void": false
   },
   "recall_vs_plain_usd_pct": -59.7313,
   "seed": 424242,
   "valid": true,
   "why": ""
  },
  {
   "deploy_after": {"at": "2026-10-03T12:48:54.464Z", "build_conclusion": "success", "build_id": "17e928e5-71cb-46aa-9592-f006f61b7588", "head": "466ebe20b9dcd717c1e6cf2459a85bcbf200d810", "worker_versions": "e19f36ae-4553-437e-9b87-b983d242976d"},
   "deploy_before": {"at": "2026-10-03T12:38:56.790Z", "build_conclusion": "success", "build_id": "17e928e5-71cb-46aa-9592-f006f61b7588", "head": "466ebe20b9dcd717c1e6cf2459a85bcbf200d810", "worker_versions": "e19f36ae-4553-437e-9b87-b983d242976d"},
   "deploy_changed": false,
   "deploy_checks": 60,
   "first": "plain",
   "pair": 3,
   "passes": true,
   "plain": {
    "cache_write_tokens": 435674,
    "client_reported_usd": 7.1166,
    "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": 0.8916},
     {"id": "c1", "usd_client_cum": 1.0386},
     {"id": "r2", "usd_client_cum": 1.983},
     {"id": "e1", "usd_client_cum": 2.1509},
     {"id": "c2", "usd_client_cum": 2.3454},
     {"id": "r3", "usd_client_cum": 3.2904},
     {"id": "e2", "usd_client_cum": 3.5138},
     {"id": "c3", "usd_client_cum": 3.7515},
     {"id": "r4", "usd_client_cum": 4.5437},
     {"id": "e3", "usd_client_cum": 4.8195},
     {"id": "t1", "usd_client_cum": 5.0997},
     {"id": "d-delimiter", "usd_client_cum": 5.2894},
     {"id": "d-date", "usd_client_cum": 5.4712},
     {"id": "t2", "usd_client_cum": 5.6723},
     {"id": "t3", "usd_client_cum": 5.8557},
     {"id": "t4", "usd_client_cum": 6.0439},
     {"id": "t5", "usd_client_cum": 6.2256},
     {"id": "q-typed-codename", "usd_client_cum": 6.3145},
     {"id": "q-typed-pager", "usd_client_cum": 6.4033},
     {"id": "q-typed-port", "usd_client_cum": 6.4921},
     {"id": "q-typed-ticket", "usd_client_cum": 6.581},
     {"id": "q-deleted-report-1", "usd_client_cum": 6.6701},
     {"id": "q-deleted-report-2", "usd_client_cum": 6.7592},
     {"id": "q-deleted-report-3", "usd_client_cum": 6.8484},
     {"id": "q-const-series-max", "usd_client_cum": 6.9376},
     {"id": "q-const-activation-bytes", "usd_client_cum": 7.0271},
     {"id": "q-const-activity-cap", "usd_client_cum": 7.1166}
    ],
    "decisions_asked": 2,
    "file": "pair-3-plain.json",
    "forwarded_tokens_max": 440319,
    "forwarded_tokens_mean": 330461,
    "partial": false,
    "probe_turns": {"n": 10, "tool_calls": 0, "usd": 0.8909, "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_net_usd": null,
    "recall_rows_by_phase": null,
    "recall_stats": null,
    "relayout_at": null,
    "relayouts": null,
    "search": 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": 7.1166,
    "usd_client": 7.1166,
    "usd_client_json": 7.1166,
    "usd_ledger": null,
    "usd_transcript": 7.1166,
    "void": false
   },
   "recall": {
    "cache_write_tokens": 190095,
    "client_reported_usd": 2.7183,
    "compactions": 0,
    "curve": [
     {"id": "s01", "usd_client_cum": 0.0524, "usd_ledger_cum": 0.0524},
     {"id": "s02", "usd_client_cum": 0.0599, "usd_ledger_cum": 0.0599},
     {"id": "s03", "usd_client_cum": 0.0672, "usd_ledger_cum": 0.0672},
     {"id": "r1", "usd_client_cum": 0.179, "usd_ledger_cum": 0.179},
     {"id": "c1", "usd_client_cum": 0.2954, "usd_ledger_cum": 0.2954},
     {"id": "r2", "usd_client_cum": 0.4335, "usd_ledger_cum": 0.4335},
     {"id": "e1", "usd_client_cum": 0.4947, "usd_ledger_cum": 0.4947},
     {"id": "c2", "usd_client_cum": 0.6207, "usd_ledger_cum": 0.6207},
     {"id": "r3", "usd_client_cum": 0.8003, "usd_ledger_cum": 0.8003},
     {"id": "e2", "usd_client_cum": 0.8696, "usd_ledger_cum": 0.8696},
     {"id": "c3", "usd_client_cum": 1.0054, "usd_ledger_cum": 1.0054},
     {"id": "r4", "usd_client_cum": 1.3168, "usd_ledger_cum": 1.3168},
     {"id": "e3", "usd_client_cum": 1.9563, "usd_ledger_cum": 1.9563},
     {"id": "t1", "usd_client_cum": 2.0487, "usd_ledger_cum": 2.0487},
     {"id": "d-delimiter", "usd_client_cum": 2.0972, "usd_ledger_cum": 2.0972},
     {"id": "d-date", "usd_client_cum": 2.1434, "usd_ledger_cum": 2.1434},
     {"id": "t2", "usd_client_cum": 2.2005, "usd_ledger_cum": 2.2005},
     {"id": "t3", "usd_client_cum": 2.2435, "usd_ledger_cum": 2.2435},
     {"id": "t4", "usd_client_cum": 2.2889, "usd_ledger_cum": 2.2889},
     {"id": "t5", "usd_client_cum": 2.3327, "usd_ledger_cum": 2.3327},
     {"id": "q-typed-codename", "usd_client_cum": 2.3533, "usd_ledger_cum": 2.3533},
     {"id": "q-typed-pager", "usd_client_cum": 2.3733, "usd_ledger_cum": 2.3733},
     {"id": "q-typed-port", "usd_client_cum": 2.3933, "usd_ledger_cum": 2.3933},
     {"id": "q-typed-ticket", "usd_client_cum": 2.4133, "usd_ledger_cum": 2.4133},
     {"id": "q-deleted-report-1", "usd_client_cum": 2.4872, "usd_ledger_cum": 2.5081},
     {"id": "q-deleted-report-2", "usd_client_cum": 2.5628, "usd_ledger_cum": 2.6086},
     {"id": "q-deleted-report-3", "usd_client_cum": 2.5857, "usd_ledger_cum": 2.6315},
     {"id": "q-const-series-max", "usd_client_cum": 2.6147, "usd_ledger_cum": 2.6814},
     {"id": "q-const-activation-bytes", "usd_client_cum": 2.6443, "usd_ledger_cum": 2.7329},
     {"id": "q-const-activity-cap", "usd_client_cum": 2.7183, "usd_ledger_cum": 2.8985}
    ],
    "decisions_asked": 2,
    "file": "pair-3-recall.json",
    "forwarded_tokens_max": 127159,
    "forwarded_tokens_mean": 79873,
    "partial": false,
    "probe_turns": {"n": 10, "tool_calls": 0, "usd": 0.3856, "write_tokens_max": 155},
    "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": 0,
    "recall_net_usd": -0.3711,
    "recall_rows_by_phase": {"recall": 57, "recall_C": 1, "recall_age_ms": 3, "recall_arch_a": 1, "recall_archived": 1, "recall_held": 26, "recall_n": 3, "recall_read_ms": 57, "recall_read_src:store": 57, "recall_relayout": 1, "recall_trigger:first": 1, "recall_trigger:ski": 2},
    "recall_stats": {"archive.bytes": 1566972, "archive.episodes": 98, "median_recall_age_ms": 449625, "reask": 1, "recall_hits": 0, "recall_judged": 0, "recalls": 18, "relayouts": 1, "saved_usd_recall": -0.371092, "triggers.first": 1, "triggers.ski": 2},
    "relayout_at": ["2026-10-03T12:43:35.861Z"],
    "relayouts": 1,
    "search": {
     "by_probe": [
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "typed", "probe": "typed-codename", "query_calls": 0},
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "typed", "probe": "typed-pager", "query_calls": 0},
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "typed", "probe": "typed-port", "query_calls": 0},
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "typed", "probe": "typed-ticket", "query_calls": 0},
      {"calls": 1, "client_saw_search": false, "docid_calls": 1, "exact": true, "found_recorded": true, "found_replay": false, "kind": "tool-output", "probe": "deleted-report-1", "query_calls": 0},
      {"calls": 1, "client_saw_search": false, "docid_calls": 1, "exact": true, "found_recorded": false, "found_replay": false, "kind": "tool-output", "probe": "deleted-report-2", "query_calls": 0},
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "tool-output", "probe": "deleted-report-3", "query_calls": 0},
      {"calls": 1, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": true, "kind": "constant", "probe": "const-series-max", "query_calls": 1},
      {"calls": 1, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": true, "kind": "constant", "probe": "const-activation-bytes", "query_calls": 1},
      {"calls": 3, "client_saw_search": false, "docid_calls": 1, "exact": true, "found_recorded": true, "found_replay": true, "kind": "constant", "probe": "const-activity-cap", "query_calls": 2}
     ],
     "exact_with_search": 5,
     "exact_without_search": 5,
     "execs_in_probes": 7,
     "execs_total": 7,
     "probes_found_recorded": 2,
     "probes_found_replay": 3,
     "probes_with_search": 5
    },
    "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": 2.8985,
    "usd_client": 2.7183,
    "usd_client_json": 2.7183,
    "usd_ledger": 2.8985,
    "usd_transcript": 2.7183,
    "void": false
   },
   "recall_vs_plain_usd_pct": -59.2713,
   "seed": 424242,
   "valid": true,
   "why": ""
  },
  {
   "deploy_after": {"at": "2026-10-03T12:57:42.702Z", "build_conclusion": "success", "build_id": "17e928e5-71cb-46aa-9592-f006f61b7588", "head": "466ebe20b9dcd717c1e6cf2459a85bcbf200d810", "worker_versions": "e19f36ae-4553-437e-9b87-b983d242976d"},
   "deploy_before": {"at": "2026-10-03T12:49:09.225Z", "build_conclusion": "success", "build_id": "17e928e5-71cb-46aa-9592-f006f61b7588", "head": "466ebe20b9dcd717c1e6cf2459a85bcbf200d810", "worker_versions": "e19f36ae-4553-437e-9b87-b983d242976d"},
   "deploy_changed": false,
   "deploy_checks": 60,
   "first": "recall",
   "pair": 4,
   "passes": false,
   "plain": {
    "cache_write_tokens": 440009,
    "client_reported_usd": 7.3357,
    "compactions": 0,
    "curve": [
     {"id": "s01", "usd_client_cum": 0.1245},
     {"id": "s02", "usd_client_cum": 0.1315},
     {"id": "s03", "usd_client_cum": 0.1373},
     {"id": "r1", "usd_client_cum": 0.8938},
     {"id": "c1", "usd_client_cum": 1.0411},
     {"id": "r2", "usd_client_cum": 1.981},
     {"id": "e1", "usd_client_cum": 2.1526},
     {"id": "c2", "usd_client_cum": 2.3474},
     {"id": "r3", "usd_client_cum": 3.2941},
     {"id": "e2", "usd_client_cum": 3.5195},
     {"id": "c3", "usd_client_cum": 3.7577},
     {"id": "r4", "usd_client_cum": 4.55},
     {"id": "e3", "usd_client_cum": 4.8279},
     {"id": "t1", "usd_client_cum": 5.1076},
     {"id": "d-delimiter", "usd_client_cum": 5.2939},
     {"id": "d-date", "usd_client_cum": 5.4782},
     {"id": "t2", "usd_client_cum": 5.6781},
     {"id": "t3", "usd_client_cum": 5.8643},
     {"id": "t4", "usd_client_cum": 6.2327},
     {"id": "t5", "usd_client_cum": 6.4169},
     {"id": "q-typed-codename", "usd_client_cum": 6.5083},
     {"id": "q-typed-pager", "usd_client_cum": 6.5998},
     {"id": "q-typed-port", "usd_client_cum": 6.6912},
     {"id": "q-typed-ticket", "usd_client_cum": 6.7827},
     {"id": "q-deleted-report-1", "usd_client_cum": 6.8746},
     {"id": "q-deleted-report-2", "usd_client_cum": 6.9666},
     {"id": "q-deleted-report-3", "usd_client_cum": 7.0586},
     {"id": "q-const-series-max", "usd_client_cum": 7.1508},
     {"id": "q-const-activation-bytes", "usd_client_cum": 7.2432},
     {"id": "q-const-activity-cap", "usd_client_cum": 7.3357}
    ],
    "decisions_asked": 2,
    "file": "pair-4-plain.json",
    "forwarded_tokens_max": 444653,
    "forwarded_tokens_mean": 335236,
    "partial": false,
    "probe_turns": {"n": 10, "tool_calls": 0, "usd": 0.9187, "write_tokens_max": 405},
    "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_net_usd": null,
    "recall_rows_by_phase": null,
    "recall_stats": null,
    "relayout_at": null,
    "relayouts": null,
    "search": 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": 7.3357,
    "usd_client": 7.3357,
    "usd_client_json": 7.3357,
    "usd_ledger": null,
    "usd_transcript": 7.3357,
    "void": false
   },
   "recall": {
    "cache_write_tokens": 180597,
    "client_reported_usd": 2.7188,
    "compactions": 0,
    "curve": [
     {"id": "s01", "usd_client_cum": 0.0358, "usd_ledger_cum": 0.0358},
     {"id": "s02", "usd_client_cum": 0.0446, "usd_ledger_cum": 0.0446},
     {"id": "s03", "usd_client_cum": 0.0672, "usd_ledger_cum": 0.0672},
     {"id": "r1", "usd_client_cum": 0.1774, "usd_ledger_cum": 0.1774},
     {"id": "c1", "usd_client_cum": 0.2925, "usd_ledger_cum": 0.2925},
     {"id": "r2", "usd_client_cum": 0.4294, "usd_ledger_cum": 0.4294},
     {"id": "e1", "usd_client_cum": 0.4909, "usd_ledger_cum": 0.4909},
     {"id": "c2", "usd_client_cum": 0.6159, "usd_ledger_cum": 0.6159},
     {"id": "r3", "usd_client_cum": 0.8662, "usd_ledger_cum": 0.8662},
     {"id": "e2", "usd_client_cum": 0.9368, "usd_ledger_cum": 0.9368},
     {"id": "c3", "usd_client_cum": 1.0723, "usd_ledger_cum": 1.0723},
     {"id": "r4", "usd_client_cum": 1.3504, "usd_ledger_cum": 1.3504},
     {"id": "e3", "usd_client_cum": 1.9073, "usd_ledger_cum": 1.9073},
     {"id": "t1", "usd_client_cum": 1.9784, "usd_ledger_cum": 1.9784},
     {"id": "d-delimiter", "usd_client_cum": 2.0404, "usd_ledger_cum": 2.0404},
     {"id": "d-date", "usd_client_cum": 2.0846, "usd_ledger_cum": 2.0846},
     {"id": "t2", "usd_client_cum": 2.1381, "usd_ledger_cum": 2.1381},
     {"id": "t3", "usd_client_cum": 2.1753, "usd_ledger_cum": 2.1753},
     {"id": "t4", "usd_client_cum": 2.2155, "usd_ledger_cum": 2.2155},
     {"id": "t5", "usd_client_cum": 2.2519, "usd_ledger_cum": 2.2519},
     {"id": "q-typed-codename", "usd_client_cum": 2.269, "usd_ledger_cum": 2.269},
     {"id": "q-typed-pager", "usd_client_cum": 2.2861, "usd_ledger_cum": 2.2861},
     {"id": "q-typed-port", "usd_client_cum": 2.3031, "usd_ledger_cum": 2.3031},
     {"id": "q-typed-ticket", "usd_client_cum": 2.3202, "usd_ledger_cum": 2.3202},
     {"id": "q-deleted-report-1", "usd_client_cum": 2.3911, "usd_ledger_cum": 2.4091},
     {"id": "q-deleted-report-2", "usd_client_cum": 2.5192, "usd_ledger_cum": 2.5569},
     {"id": "q-deleted-report-3", "usd_client_cum": 2.5381, "usd_ledger_cum": 2.5758},
     {"id": "q-const-series-max", "usd_client_cum": 2.5637, "usd_ledger_cum": 2.6201},
     {"id": "q-const-activation-bytes", "usd_client_cum": 2.5953, "usd_ledger_cum": 2.67},
     {"id": "q-const-activity-cap", "usd_client_cum": 2.7188, "usd_ledger_cum": 2.8395}
    ],
    "decisions_asked": 2,
    "file": "pair-4-recall.json",
    "forwarded_tokens_max": 128152,
    "forwarded_tokens_mean": 73663,
    "partial": false,
    "probe_turns": {"n": 10, "tool_calls": 0, "usd": 0.467, "write_tokens_max": 301},
    "probes": 8,
    "probes_by_kind": [
     {"exact": 4, "kind": "typed", "of": 4},
     {"exact": 2, "kind": "tool-output", "of": 3},
     {"exact": 2, "kind": "constant", "of": 3}
    ],
    "probes_contains": 10,
    "probes_of": 10,
    "reasks": 0,
    "recall_fail_rows": 0,
    "recall_net_usd": -0.1767,
    "recall_rows_by_phase": {"recall": 60, "recall_C": 1, "recall_age_ms": 3, "recall_arch_a": 1, "recall_archived": 1, "recall_held": 28, "recall_n": 3, "recall_read_ms": 60, "recall_read_src:store": 60, "recall_relayout": 1, "recall_trigger:first": 1, "recall_trigger:ski": 2},
    "recall_stats": {"archive.bytes": 1558976, "archive.episodes": 97, "median_recall_age_ms": 379840, "reask": 0, "recall_hits": 0, "recall_judged": 0, "recalls": 18, "relayouts": 1, "saved_usd_recall": -0.176683, "triggers.first": 1, "triggers.ski": 2},
    "relayout_at": ["2026-10-03T12:53:02.858Z"],
    "relayouts": 1,
    "search": {
     "by_probe": [
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "typed", "probe": "typed-codename", "query_calls": 0},
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "typed", "probe": "typed-pager", "query_calls": 0},
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "typed", "probe": "typed-port", "query_calls": 0},
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "typed", "probe": "typed-ticket", "query_calls": 0},
      {"calls": 1, "client_saw_search": false, "docid_calls": 1, "exact": true, "found_recorded": true, "found_replay": false, "kind": "tool-output", "probe": "deleted-report-1", "query_calls": 0},
      {"calls": 2, "client_saw_search": false, "docid_calls": 2, "exact": false, "found_recorded": true, "found_replay": false, "kind": "tool-output", "probe": "deleted-report-2", "query_calls": 0},
      {"calls": 0, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": false, "kind": "tool-output", "probe": "deleted-report-3", "query_calls": 0},
      {"calls": 1, "client_saw_search": false, "docid_calls": 0, "exact": true, "found_recorded": false, "found_replay": true, "kind": "constant", "probe": "const-series-max", "query_calls": 1},
      {"calls": 1, "client_saw_search": false, "docid_calls": 0, "exact": false, "found_recorded": true, "found_replay": true, "kind": "constant", "probe": "const-activation-bytes", "query_calls": 1},
      {"calls": 2, "client_saw_search": false, "docid_calls": 1, "exact": true, "found_recorded": true, "found_replay": true, "kind": "constant", "probe": "const-activity-cap", "query_calls": 1}
     ],
     "exact_with_search": 3,
     "exact_without_search": 5,
     "execs_in_probes": 7,
     "execs_total": 7,
     "probes_found_recorded": 4,
     "probes_found_replay": 3,
     "probes_with_search": 5
    },
    "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": 2.8395,
    "usd_client": 2.7188,
    "usd_client_json": 2.7188,
    "usd_ledger": 2.8395,
    "usd_transcript": 2.7188,
    "void": false
   },
   "recall_vs_plain_usd_pct": -61.292,
   "seed": 424242,
   "valid": true,
   "why": ""
  }
 ],
 "pairs_passing": 3,
 "pairs_planned": 4,
 "pairs_run": 4,
 "pairs_valid": 4,
 "prereg": "pre-registered in gateway commit 5e8b5c85 (2026-10-03T15:07:40+03:00), before the first pair",
 "prereg_check": {"commit": "5e8b5c85", "committed": "2026-10-03T12:07:40.000Z", "plan": "prereg/long-session-amendment-4-prereg.md", "run_started": "2026-10-03T12:20:25.934Z", "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": {"dry4-plain": 0.9044, "dry4-recall-b4k": 0.4951, "p1-plain": 7.1465, "p1-recall": 3.1708, "p2-plain": 7.3454, "p2-recall": 2.9579, "p3-plain": 7.1166, "p3-recall": 2.8985, "p4-plain": 7.3357, "p4-recall": 2.8395}
 },
 "withheld_fields": {"gateway_without_recall_usd_est": "a worked-out figure for the session without Recall, not a measurement"}
}