{
  "knowledge": [
    {
      "evidence_id": "m_c558d1541c4b9d0f",
      "text": "2026-08-01. Mara started project Larch. They chose BATCH-0-741 for batch delivery. The initial retry policy was RETRY-0-108.",
      "timestamp": "2026-08-01T12:00:00Z"
    },
    {
      "evidence_id": "m_d7a9d1f22107f233",
      "text": "2026-08-08. Mara reported that BATCH-0-741 lost updates during GATE-0-519 outages. The team tested REPLAY-0-632, which restored the missing updates. No change of delivery mode was approved at this meeting.",
      "timestamp": "2026-08-08T12:00:00Z"
    },
    {
      "evidence_id": "m_eb85892166acc8ef",
      "text": "2026-08-01. Ivo started project Cobalt. They chose BATCH-1-741 for batch delivery. The initial retry policy was RETRY-1-108.",
      "timestamp": "2026-08-01T12:00:00Z"
    },
    {
      "evidence_id": "m_f7ff6441bdd98e53",
      "text": "2026-08-08. Ivo reported that BATCH-1-741 lost updates during GATE-1-519 outages. The team tested REPLAY-1-632, which restored the missing updates. No change of delivery mode was approved at this meeting.",
      "timestamp": "2026-08-08T12:00:00Z"
    },
    {
      "evidence_id": "e_6d31f2e707b427b03ebe5cab",
      "text": "Kestrel-0 signed return policy. Ordinary goods have a 7-day return window. The complete exception list consists of two clauses: custom-made goods use exception CUSTOM-0-7c2; quality defects use exception DEFECT-0-9b4 and are exempt from the 7-day limit. Both exception identifiers must be retained when listing all exceptions. The old informal draft listed only DRAFT-0-2d1 and has no authority."
    },
    {
      "evidence_id": "e_44d1d038212b769f1cefdbec",
      "text": "Kestrel-0 approved regional routing table. Columns are region, response window, queue.\nNorth | 4 hours | NORTH-0-6a8\nSouth | 12 hours | SOUTH-0-3e5\nRows are independent; do not mix a region with another row."
    }
  ],
  "memory_initial": [
    {
      "evidence_id": "m_c558d1541c4b9d0f",
      "text": "2026-08-01. Mara started project Larch. They chose BATCH-0-741 for batch delivery. The initial retry policy was RETRY-0-108.",
      "timestamp": "2026-08-01T12:00:00Z"
    },
    {
      "evidence_id": "m_d7a9d1f22107f233",
      "text": "2026-08-08. Mara reported that BATCH-0-741 lost updates during GATE-0-519 outages. The team tested REPLAY-0-632, which restored the missing updates. No change of delivery mode was approved at this meeting.",
      "timestamp": "2026-08-08T12:00:00Z"
    },
    {
      "evidence_id": "m_eb85892166acc8ef",
      "text": "2026-08-01. Ivo started project Cobalt. They chose BATCH-1-741 for batch delivery. The initial retry policy was RETRY-1-108.",
      "timestamp": "2026-08-01T12:00:00Z"
    },
    {
      "evidence_id": "m_f7ff6441bdd98e53",
      "text": "2026-08-08. Ivo reported that BATCH-1-741 lost updates during GATE-1-519 outages. The team tested REPLAY-1-632, which restored the missing updates. No change of delivery mode was approved at this meeting.",
      "timestamp": "2026-08-08T12:00:00Z"
    }
  ],
  "memory_updates": [
    {
      "evidence_id": "m_99ab4da076b7c734",
      "text": "2026-08-15. Mara approved switching Larch from BATCH-0-741 to STREAM-0-286. This supersedes the August 1 delivery decision. The cause was the GATE-0-519 outages recorded on August 8; keep the successful recovery procedure from that incident.",
      "timestamp": "2026-08-15T12:00:00Z"
    },
    {
      "evidence_id": "m_d6297f549707f98d",
      "text": "2026-08-22. Mara reviewed Larch. Keep the August 15 delivery choice. For the next outage, use the recovery procedure validated on August 8. The retry policy is now RETRY-0-907, superseding RETRY-0-108. No budget owner has been appointed.",
      "timestamp": "2026-08-22T12:00:00Z"
    },
    {
      "evidence_id": "m_70bd88efee712a9a",
      "text": "2026-08-15. Ivo approved switching Cobalt from BATCH-1-741 to STREAM-1-286. This supersedes the August 1 delivery decision. The cause was the GATE-1-519 outages recorded on August 8; keep the successful recovery procedure from that incident.",
      "timestamp": "2026-08-15T12:00:00Z"
    },
    {
      "evidence_id": "m_4a54af2b364be476",
      "text": "2026-08-22. Ivo reviewed Cobalt. Keep the August 15 delivery choice. For the next outage, use the recovery procedure validated on August 8. The retry policy is now RETRY-1-907, superseding RETRY-1-108. No budget owner has been appointed.",
      "timestamp": "2026-08-22T12:00:00Z"
    }
  ],
  "pre_update_queries": [
    {
      "case_id": "memory-0-2",
      "question": "What delivery mode was still approved for Mara's Larch project immediately after the August 8 meeting?",
      "lane": "temporal",
      "split": "test"
    },
    {
      "case_id": "memory-1-2",
      "question": "What delivery mode was still approved for Ivo's Cobalt project immediately after the August 8 meeting?",
      "lane": "temporal",
      "split": "test"
    }
  ],
  "post_update_queries": [
    {
      "case_id": "memory-0-0",
      "question": "As of August 22, what delivery mode and retry policy apply to Mara's Larch project?",
      "lane": "updates",
      "split": "test"
    },
    {
      "case_id": "memory-0-1",
      "question": "For the next outage on Larch, which recovery procedure should Mara use, and which outage gate led to the delivery-mode change?",
      "lane": "cross_session",
      "split": "test"
    },
    {
      "case_id": "memory-0-2",
      "question": "What delivery mode was still approved for Mara's Larch project immediately after the August 8 meeting?",
      "lane": "temporal",
      "split": "test"
    },
    {
      "case_id": "memory-0-3",
      "question": "Who is the appointed budget owner for Mara's Larch project as of August 22?",
      "lane": "abstention",
      "split": "test"
    },
    {
      "case_id": "memory-1-0",
      "question": "As of August 22, what delivery mode and retry policy apply to Ivo's Cobalt project?",
      "lane": "updates",
      "split": "test"
    },
    {
      "case_id": "memory-1-1",
      "question": "For the next outage on Cobalt, which recovery procedure should Ivo use, and which outage gate led to the delivery-mode change?",
      "lane": "cross_session",
      "split": "test"
    },
    {
      "case_id": "memory-1-2",
      "question": "What delivery mode was still approved for Ivo's Cobalt project immediately after the August 8 meeting?",
      "lane": "temporal",
      "split": "test"
    },
    {
      "case_id": "memory-1-3",
      "question": "Who is the appointed budget owner for Ivo's Cobalt project as of August 22?",
      "lane": "abstention",
      "split": "test"
    },
    {
      "action": "query",
      "scope": "documents",
      "case_id": "doc-000",
      "lane": "document_exceptions",
      "question": "List both exception identifiers in the signed Kestrel-0 return policy."
    },
    {
      "action": "query",
      "scope": "documents",
      "case_id": "doc-001",
      "lane": "document_table",
      "question": "What is the approved queue for the South region of Kestrel-0?"
    },
    {
      "case_id": "combined-001",
      "question": "As of August 22, what delivery mode and retry policy apply to Mara's Larch project? Also list both exception identifiers in the signed Kestrel-0 return policy.",
      "lane": "combined_knowledge_memory",
      "split": "test"
    }
  ],
  "calibration": {
    "memory": [
      {
        "evidence_id": "m_11ab570fcca41f15",
        "text": "2026-08-01. Devon started project Cedar. They chose BATCH-4-741 for batch delivery. The initial retry policy was RETRY-4-108.",
        "timestamp": "2026-08-01T12:00:00Z"
      }
    ],
    "knowledge": [
      {
        "evidence_id": "e_be0f90d28465e92fcac2023d",
        "text": "Kestrel-3 signed return policy. Ordinary goods have a 7-day return window. The complete exception list consists of two clauses: custom-made goods use exception CUSTOM-3-7c2; quality defects use exception DEFECT-3-9b4 and are exempt from the 7-day limit. Both exception identifiers must be retained when listing all exceptions. The old informal draft listed only DRAFT-3-2d1 and has no authority."
      }
    ],
    "question": "Which initial delivery mode did Devon choose for Cedar, and what are both Kestrel-3 return exception identifiers?"
  }
}