{
  "version": "2026-03-12",
  "scope": "Validated public and first-party data requirements for OSUN benchmark tasks.",
  "task_ids": [
    "fermentos_yield",
    "farmos_output",
    "wastezero_covers",
    "sourcegrid_leadtime",
    "autokitchen_throughput",
    "ghostflow_dispatch",
    "rentgrid_occupancy",
    "recipeproof_compliance",
    "localsource_packaging",
    "localsource_ingredients"
  ],
  "sources": [
    {
      "source_id": "usda_nass_quickstats",
      "label": "USDA NASS Quick Stats API",
      "provider": "USDA NASS",
      "kind": "government_primary",
      "url": "https://www.nass.usda.gov/developer/",
      "corpus_refs": [
        "C6"
      ]
    },
    {
      "source_id": "usda_ams_local_food",
      "label": "USDA AMS Local Food Directories API",
      "provider": "USDA AMS",
      "kind": "government_primary",
      "url": "https://www.ams.usda.gov/local-food-directories",
      "corpus_refs": [
        "C16"
      ]
    },
    {
      "source_id": "usda_ers_food_markets",
      "label": "USDA ERS Food Markets & Expenditures",
      "provider": "USDA ERS",
      "kind": "government_primary",
      "url": "https://www.ers.usda.gov/data-products/",
      "corpus_refs": [
        "C11"
      ]
    },
    {
      "source_id": "usda_fooddata_central",
      "label": "USDA FoodData Central API",
      "provider": "USDA",
      "kind": "government_primary",
      "url": "https://fdc.nal.usda.gov/api-guide/",
      "corpus_refs": [
        "C8"
      ]
    },
    {
      "source_id": "noaa_cdo",
      "label": "NOAA Climate Data Online API",
      "provider": "NOAA NCEI",
      "kind": "government_primary",
      "url": "https://www.ncei.noaa.gov/cdo-web/webservices/v2",
      "corpus_refs": [
        "C6"
      ]
    },
    {
      "source_id": "nasa_power",
      "label": "NASA POWER API",
      "provider": "NASA",
      "kind": "government_primary",
      "url": "https://power.larc.nasa.gov/docs/services/api/",
      "corpus_refs": [
        "C6"
      ]
    },
    {
      "source_id": "eia_v2",
      "label": "U.S. Energy Information Administration API",
      "provider": "EIA",
      "kind": "government_primary",
      "url": "https://www.eia.gov/opendata/",
      "corpus_refs": [
        "C7"
      ]
    },
    {
      "source_id": "bls_public_data",
      "label": "BLS Public Data API",
      "provider": "U.S. Bureau of Labor Statistics",
      "kind": "government_primary",
      "url": "https://www.bls.gov/developers/",
      "corpus_refs": [
        "C2"
      ]
    },
    {
      "source_id": "census_data_api",
      "label": "U.S. Census Data API",
      "provider": "U.S. Census Bureau",
      "kind": "government_primary",
      "url": "https://www.census.gov/data/developers/data-sets.html",
      "corpus_refs": [
        "C2"
      ]
    },
    {
      "source_id": "fda_food_additive_status",
      "label": "FDA Food Additive Status List",
      "provider": "FDA",
      "kind": "government_primary",
      "url": "https://www.fda.gov/food/food-additives-petitions/food-additive-status-list",
      "corpus_refs": [
        "C14"
      ]
    },
    {
      "source_id": "fda_fcs_inventory",
      "label": "FDA Food Contact Substance Inventory",
      "provider": "FDA",
      "kind": "government_primary",
      "url": "https://www.fda.gov/food/packaging-food-contact-substances-fcs/inventory-effective-food-contact-substance-fcs-notifications",
      "corpus_refs": [
        "C17"
      ]
    },
    {
      "source_id": "fda_enforcement_report",
      "label": "FDA Enforcement Reports",
      "provider": "FDA",
      "kind": "government_primary",
      "url": "https://www.fda.gov/safety/enforcement-reports",
      "corpus_refs": [
        "C14"
      ]
    },
    {
      "source_id": "first_party_runtime_telemetry",
      "label": "OSUN First-Party Runtime Telemetry",
      "provider": "OSUN Platform",
      "kind": "internal_telemetry",
      "url": "https://osunandco.com/",
      "corpus_refs": [
        "C3",
        "C5"
      ]
    }
  ],
  "requirements": [
    {
      "requirement_id": "fermentos_yield_inputs",
      "suite": "agtech",
      "task_ids": [
        "fermentos_yield"
      ],
      "source_ids": [
        "usda_nass_quickstats",
        "noaa_cdo",
        "nasa_power",
        "eia_v2",
        "first_party_runtime_telemetry"
      ],
      "dataset_contract": {
        "cadence": "daily",
        "freshness_max_days": 7,
        "minimum_rows": 365,
        "max_null_ratio": 0.02,
        "required_fields": [
          {
            "name": "timestamp_utc",
            "type": "datetime"
          },
          {
            "name": "yield_index",
            "type": "number"
          },
          {
            "name": "energy_cost_usd_kwh",
            "type": "number"
          },
          {
            "name": "cooling_degree_days",
            "type": "number"
          },
          {
            "name": "feedstock_availability_idx",
            "type": "number"
          }
        ],
        "numeric_bounds": {
          "yield_index": {
            "min": 0.0,
            "max": 100000.0
          },
          "energy_cost_usd_kwh": {
            "min": 0.0,
            "max": 5.0
          },
          "cooling_degree_days": {
            "min": 0.0,
            "max": 250.0
          }
        }
      }
    },
    {
      "requirement_id": "farmos_output_inputs",
      "suite": "agtech",
      "task_ids": [
        "farmos_output"
      ],
      "source_ids": [
        "usda_nass_quickstats",
        "noaa_cdo",
        "nasa_power",
        "census_data_api",
        "first_party_runtime_telemetry"
      ],
      "dataset_contract": {
        "cadence": "weekly",
        "freshness_max_days": 10,
        "minimum_rows": 260,
        "max_null_ratio": 0.03,
        "required_fields": [
          {
            "name": "week_start",
            "type": "date"
          },
          {
            "name": "output_tonnes",
            "type": "number"
          },
          {
            "name": "soil_moisture_pct",
            "type": "number"
          },
          {
            "name": "precip_mm",
            "type": "number"
          },
          {
            "name": "available_labor_hours",
            "type": "number"
          }
        ],
        "numeric_bounds": {
          "output_tonnes": {
            "min": 0.0,
            "max": 1000000.0
          },
          "soil_moisture_pct": {
            "min": 0.0,
            "max": 100.0
          }
        }
      }
    },
    {
      "requirement_id": "wastezero_covers_inputs",
      "suite": "foodtech",
      "task_ids": [
        "wastezero_covers"
      ],
      "source_ids": [
        "usda_ers_food_markets",
        "bls_public_data",
        "first_party_runtime_telemetry"
      ],
      "dataset_contract": {
        "cadence": "daily",
        "freshness_max_days": 3,
        "minimum_rows": 540,
        "max_null_ratio": 0.02,
        "required_fields": [
          {
            "name": "service_date",
            "type": "date"
          },
          {
            "name": "covers",
            "type": "integer"
          },
          {
            "name": "waste_kg",
            "type": "number"
          },
          {
            "name": "menu_price_index",
            "type": "number"
          },
          {
            "name": "promo_intensity",
            "type": "number"
          }
        ],
        "numeric_bounds": {
          "covers": {
            "min": 0.0,
            "max": 500000.0
          },
          "waste_kg": {
            "min": 0.0,
            "max": 100000.0
          }
        }
      }
    },
    {
      "requirement_id": "sourcegrid_leadtime_inputs",
      "suite": "foodtech",
      "task_ids": [
        "sourcegrid_leadtime"
      ],
      "source_ids": [
        "usda_ams_local_food",
        "census_data_api",
        "eia_v2",
        "first_party_runtime_telemetry"
      ],
      "dataset_contract": {
        "cadence": "daily",
        "freshness_max_days": 7,
        "minimum_rows": 400,
        "max_null_ratio": 0.03,
        "required_fields": [
          {
            "name": "date",
            "type": "date"
          },
          {
            "name": "supplier_id",
            "type": "string"
          },
          {
            "name": "lead_time_days",
            "type": "number"
          },
          {
            "name": "distance_km",
            "type": "number"
          },
          {
            "name": "fuel_cost_index",
            "type": "number"
          }
        ],
        "numeric_bounds": {
          "lead_time_days": {
            "min": 0.0,
            "max": 180.0
          },
          "distance_km": {
            "min": 0.0,
            "max": 20000.0
          }
        }
      }
    },
    {
      "requirement_id": "autokitchen_throughput_inputs",
      "suite": "restaurantops",
      "task_ids": [
        "autokitchen_throughput"
      ],
      "source_ids": [
        "bls_public_data",
        "census_data_api",
        "first_party_runtime_telemetry"
      ],
      "dataset_contract": {
        "cadence": "hourly",
        "freshness_max_days": 2,
        "minimum_rows": 720,
        "max_null_ratio": 0.01,
        "required_fields": [
          {
            "name": "timestamp_utc",
            "type": "datetime"
          },
          {
            "name": "orders_completed",
            "type": "integer"
          },
          {
            "name": "station_utilization_pct",
            "type": "number"
          },
          {
            "name": "queue_depth",
            "type": "integer"
          },
          {
            "name": "staffed_stations",
            "type": "integer"
          }
        ],
        "numeric_bounds": {
          "orders_completed": {
            "min": 0.0,
            "max": 100000.0
          },
          "station_utilization_pct": {
            "min": 0.0,
            "max": 100.0
          }
        }
      }
    },
    {
      "requirement_id": "ghostflow_dispatch_inputs",
      "suite": "restaurantops",
      "task_ids": [
        "ghostflow_dispatch"
      ],
      "source_ids": [
        "noaa_cdo",
        "eia_v2",
        "first_party_runtime_telemetry"
      ],
      "dataset_contract": {
        "cadence": "hourly",
        "freshness_max_days": 2,
        "minimum_rows": 720,
        "max_null_ratio": 0.015,
        "required_fields": [
          {
            "name": "timestamp_utc",
            "type": "datetime"
          },
          {
            "name": "dispatch_delay_minutes",
            "type": "number"
          },
          {
            "name": "order_volume",
            "type": "integer"
          },
          {
            "name": "weather_alert_index",
            "type": "number"
          },
          {
            "name": "fuel_cost_index",
            "type": "number"
          }
        ],
        "numeric_bounds": {
          "dispatch_delay_minutes": {
            "min": 0.0,
            "max": 720.0
          },
          "order_volume": {
            "min": 0.0,
            "max": 500000.0
          }
        }
      }
    },
    {
      "requirement_id": "rentgrid_occupancy_inputs",
      "suite": "restaurantops",
      "task_ids": [
        "rentgrid_occupancy"
      ],
      "source_ids": [
        "census_data_api",
        "bls_public_data",
        "first_party_runtime_telemetry"
      ],
      "dataset_contract": {
        "cadence": "daily",
        "freshness_max_days": 7,
        "minimum_rows": 365,
        "max_null_ratio": 0.02,
        "required_fields": [
          {
            "name": "date",
            "type": "date"
          },
          {
            "name": "occupied_slots",
            "type": "integer"
          },
          {
            "name": "available_slots",
            "type": "integer"
          },
          {
            "name": "avg_hourly_rate_usd",
            "type": "number"
          },
          {
            "name": "market_demand_index",
            "type": "number"
          }
        ],
        "numeric_bounds": {
          "occupied_slots": {
            "min": 0.0,
            "max": 100000.0
          },
          "avg_hourly_rate_usd": {
            "min": 0.0,
            "max": 10000.0
          }
        }
      }
    },
    {
      "requirement_id": "recipeproof_compliance_inputs",
      "suite": "foodtech",
      "task_ids": [
        "recipeproof_compliance"
      ],
      "source_ids": [
        "fda_food_additive_status",
        "fda_fcs_inventory",
        "fda_enforcement_report",
        "usda_fooddata_central",
        "first_party_runtime_telemetry"
      ],
      "dataset_contract": {
        "cadence": "daily",
        "freshness_max_days": 5,
        "minimum_rows": 220,
        "max_null_ratio": 0.01,
        "required_fields": [
          {
            "name": "record_date",
            "type": "date"
          },
          {
            "name": "ingredient_identifier",
            "type": "string"
          },
          {
            "name": "status_flag",
            "type": "string"
          },
          {
            "name": "packaging_class",
            "type": "string"
          },
          {
            "name": "compliance_risk_score",
            "type": "number"
          }
        ],
        "numeric_bounds": {
          "compliance_risk_score": {
            "min": 0.0,
            "max": 1.0
          }
        }
      }
    },
    {
      "requirement_id": "localsource_packaging_inputs",
      "suite": "localsource",
      "task_ids": [
        "localsource_packaging"
      ],
      "source_ids": [
        "usda_ams_local_food",
        "fda_fcs_inventory",
        "eia_v2",
        "first_party_runtime_telemetry"
      ],
      "dataset_contract": {
        "cadence": "daily",
        "freshness_max_days": 5,
        "minimum_rows": 260,
        "max_null_ratio": 0.02,
        "required_fields": [
          {
            "name": "record_date",
            "type": "date"
          },
          {
            "name": "supplier_id",
            "type": "string"
          },
          {
            "name": "packaging_unit_cost_usd",
            "type": "number"
          },
          {
            "name": "transport_cost_usd",
            "type": "number"
          },
          {
            "name": "fcs_approved",
            "type": "boolean"
          }
        ],
        "numeric_bounds": {
          "packaging_unit_cost_usd": {
            "min": 0.0,
            "max": 10000.0
          },
          "transport_cost_usd": {
            "min": 0.0,
            "max": 100000.0
          }
        }
      }
    },
    {
      "requirement_id": "localsource_ingredients_inputs",
      "suite": "localsource",
      "task_ids": [
        "localsource_ingredients"
      ],
      "source_ids": [
        "usda_ams_local_food",
        "usda_fooddata_central",
        "usda_ers_food_markets",
        "noaa_cdo",
        "first_party_runtime_telemetry"
      ],
      "dataset_contract": {
        "cadence": "daily",
        "freshness_max_days": 3,
        "minimum_rows": 320,
        "max_null_ratio": 0.02,
        "required_fields": [
          {
            "name": "record_date",
            "type": "date"
          },
          {
            "name": "ingredient_code",
            "type": "string"
          },
          {
            "name": "local_unit_cost_usd",
            "type": "number"
          },
          {
            "name": "distance_km",
            "type": "number"
          },
          {
            "name": "freshness_hours",
            "type": "number"
          }
        ],
        "numeric_bounds": {
          "local_unit_cost_usd": {
            "min": 0.0,
            "max": 10000.0
          },
          "freshness_hours": {
            "min": 0.0,
            "max": 720.0
          }
        }
      }
    }
  ]
}