{
  "snapshot_type": "rl_trajectory_snapshot",
  "gamma": 0.9,
  "episodes": [
    {
      "episode_id": "case_001",
      "return": 1.6,
      "steps": [
        {
          "t": 0,
          "state_id": "ticket_nuevo",
          "action": "pedir_dato",
          "action_probability": 0.34,
          "reward": -0.2,
          "next_state_id": "esperando_usuario",
          "terminal": false,
          "policy_version": "2026-06-07.3",
          "reward_version": "support_reward.v2",
          "trace_id": "trace_case001_0"
        },
        {
          "t": 1,
          "state_id": "ticket_con_evidencia",
          "action": "responder_con_cita",
          "action_probability": 0.72,
          "reward": 2.0,
          "next_state_id": "ticket_resuelto",
          "terminal": true,
          "policy_version": "2026-06-07.3",
          "reward_version": "support_reward.v2",
          "trace_id": "trace_case001_1"
        }
      ]
    },
    {
      "episode_id": "case_002",
      "return": 0.26,
      "steps": [
        {
          "t": 0,
          "state_id": "ticket_nuevo",
          "action": "consultar_rag",
          "action_probability": 0.43,
          "reward": -0.1,
          "next_state_id": "ticket_con_evidencia",
          "terminal": false,
          "policy_version": "2026-06-07.3",
          "reward_version": "support_reward.v2",
          "trace_id": "trace_case002_0"
        },
        {
          "t": 1,
          "state_id": "ticket_con_evidencia",
          "action": "escalar_revision",
          "action_probability": 0.18,
          "reward": 0.4,
          "next_state_id": "revision_humana",
          "terminal": true,
          "policy_version": "2026-06-07.3",
          "reward_version": "support_reward.v2",
          "trace_id": "trace_case002_1"
        }
      ]
    },
    {
      "episode_id": "case_003",
      "return": 1.79,
      "steps": [
        {
          "t": 0,
          "state_id": "ticket_nuevo",
          "action": "consultar_rag",
          "action_probability": 0.39,
          "reward": -0.1,
          "next_state_id": "ticket_con_evidencia",
          "terminal": false,
          "policy_version": "2026-06-07.3",
          "reward_version": "support_reward.v2",
          "trace_id": "trace_case003_0"
        },
        {
          "t": 1,
          "state_id": "ticket_con_evidencia",
          "action": "responder_con_cita",
          "action_probability": 0.76,
          "reward": 2.1,
          "next_state_id": "ticket_resuelto",
          "terminal": true,
          "policy_version": "2026-06-07.3",
          "reward_version": "support_reward.v2",
          "trace_id": "trace_case003_1"
        }
      ]
    },
    {
      "episode_id": "case_004",
      "return": -1.46,
      "steps": [
        {
          "t": 0,
          "state_id": "ticket_nuevo",
          "action": "pedir_dato",
          "action_probability": 0.31,
          "reward": -0.2,
          "next_state_id": "esperando_usuario",
          "terminal": false,
          "policy_version": "2026-06-07.3",
          "reward_version": "support_reward.v2",
          "trace_id": "trace_case004_0"
        },
        {
          "t": 1,
          "state_id": "ticket_con_evidencia",
          "action": "responder",
          "action_probability": 0.22,
          "reward": -1.4,
          "next_state_id": "ticket_reabierto",
          "terminal": true,
          "policy_version": "2026-06-07.3",
          "reward_version": "support_reward.v2",
          "trace_id": "trace_case004_1"
        }
      ]
    }
  ],
  "snapshot_hash": "42fef8c39c537ea134973d404256db78ab515cd185c82db641b8b42673508194"
}
