{
  "$defs": {
    "CategoryResult": {
      "description": "Result for one taxonomy category evaluated by a guardrail.\n\nCategories give multi-label and taxonomy guardrails (DuoGuard, Llama Guard\nS-codes, Azure severities, binary classifiers' label distributions) a\nlossless, uniform home instead of overloading ``explanation``.\n\nExample:\n    >>> CategoryResult(name=\"S1\", description=\"Violent Crimes\", triggered=True)",
      "properties": {
        "description": {
          "anyOf": [
            {
              "type": "string"
            },
            {
              "type": "null"
            }
          ],
          "default": null,
          "title": "Description"
        },
        "name": {
          "title": "Name",
          "type": "string"
        },
        "score": {
          "anyOf": [
            {
              "type": "number"
            },
            {
              "type": "null"
            }
          ],
          "default": null,
          "title": "Score"
        },
        "severity": {
          "anyOf": [
            {
              "type": "integer"
            },
            {
              "type": "null"
            }
          ],
          "default": null,
          "title": "Severity"
        },
        "triggered": {
          "anyOf": [
            {
              "type": "boolean"
            },
            {
              "type": "null"
            }
          ],
          "default": null,
          "title": "Triggered"
        }
      },
      "required": [
        "name"
      ],
      "title": "CategoryResult",
      "type": "object"
    },
    "GuardrailUsage": {
      "description": "Provenance and cost envelope for one ``validate()`` call.",
      "properties": {
        "completion_tokens": {
          "anyOf": [
            {
              "type": "integer"
            },
            {
              "type": "null"
            }
          ],
          "default": null,
          "title": "Completion Tokens"
        },
        "latency_ms": {
          "anyOf": [
            {
              "type": "number"
            },
            {
              "type": "null"
            }
          ],
          "default": null,
          "title": "Latency Ms"
        },
        "model_id": {
          "anyOf": [
            {
              "type": "string"
            },
            {
              "type": "null"
            }
          ],
          "default": null,
          "title": "Model Id"
        },
        "prompt_tokens": {
          "anyOf": [
            {
              "type": "integer"
            },
            {
              "type": "null"
            }
          ],
          "default": null,
          "title": "Prompt Tokens"
        }
      },
      "title": "GuardrailUsage",
      "type": "object"
    },
    "SpanResult": {
      "description": "A character span flagged by a guardrail.\n\nReserved for span-producing guardrails (PII/NER detection, span-level\ntoxicity, token classification). Offsets are zero-based character indices\ninto the validated text.",
      "properties": {
        "end": {
          "title": "End",
          "type": "integer"
        },
        "label": {
          "anyOf": [
            {
              "type": "string"
            },
            {
              "type": "null"
            }
          ],
          "default": null,
          "title": "Label"
        },
        "score": {
          "anyOf": [
            {
              "type": "number"
            },
            {
              "type": "null"
            }
          ],
          "default": null,
          "title": "Score"
        },
        "start": {
          "title": "Start",
          "type": "integer"
        },
        "text": {
          "anyOf": [
            {
              "type": "string"
            },
            {
              "type": "null"
            }
          ],
          "default": null,
          "title": "Text"
        }
      },
      "required": [
        "start",
        "end"
      ],
      "title": "SpanResult",
      "type": "object"
    }
  },
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "description": "Represents the output of a guardrail evaluation with runtime validation.\n\nThis is the single concrete result shape shared by every guardrail\nimplementation, so application code can swap guardrails without changes.\n\nExample:\n    >>> output = GuardrailOutput(valid=True, explanation=\"Content is safe\", score=0.05)\n    >>> output.valid\n    True",
  "properties": {
    "action": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "title": "Action"
    },
    "categories": {
      "items": {
        "$ref": "#/$defs/CategoryResult"
      },
      "title": "Categories",
      "type": "array"
    },
    "explanation": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "title": "Explanation"
    },
    "extra": {
      "anyOf": [
        {
          "type": "object"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "title": "Extra"
    },
    "modified_text": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "title": "Modified Text"
    },
    "raw": {
      "anyOf": [
        {},
        {
          "type": "null"
        }
      ],
      "default": null,
      "title": "Raw"
    },
    "score": {
      "anyOf": [
        {
          "type": "number"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "title": "Score"
    },
    "spans": {
      "anyOf": [
        {
          "items": {
            "$ref": "#/$defs/SpanResult"
          },
          "type": "array"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "title": "Spans"
    },
    "usage": {
      "anyOf": [
        {
          "$ref": "#/$defs/GuardrailUsage"
        },
        {
          "type": "null"
        }
      ],
      "default": null
    },
    "valid": {
      "title": "Valid",
      "type": "boolean"
    }
  },
  "required": [
    "valid"
  ],
  "title": "GuardrailOutput",
  "type": "object"
}