AnalysisProvenance

Fields

Which engine actually produced this conclusion, and why.

15 fields

confidence_basis

string or null

fallback_from

string or null

fallback_occurred

boolean

fallback_reason

string or null

hint_id

string or null

hint_match_field

string or null

hint_pattern

string or null

hint_set_by

string or null

llm_model

string or null

llm_provider

string or null

mode_requested

string or null

mode_resolved

string or null

mode_used

string or null

prompt_versions

object

threshold_check

ThresholdCheck or null

Raw schema (JSON)

{
  "description": "US-15.1 — which engine actually produced this conclusion, and why.\n\nPopulated verbatim from ``AIAnalysis.routing_metadata`` (written by the\nanalysis router). It is NEVER recomputed or inferred: rows analysed before\nthis feature carry no routing metadata, so ``AnalysisResponse.provenance``\nis ``None`` for them rather than a plausible-looking guess.\n\nThe point of this block is a specific honesty case: when the LLM was\nunavailable and the rules engine ran instead, the card must be able to say\nso (``fallback_occurred`` / ``fallback_from`` / ``fallback_reason``)\ninstead of silently presenting heuristics as model output.",
  "properties": {
    "confidence_basis": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "title": "Confidence Basis"
    },
    "fallback_from": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "title": "Fallback From"
    },
    "fallback_occurred": {
      "default": false,
      "title": "Fallback Occurred",
      "type": "boolean"
    },
    "fallback_reason": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "title": "Fallback Reason"
    },
    "hint_id": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "title": "Hint Id"
    },
    "hint_match_field": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "title": "Hint Match Field"
    },
    "hint_pattern": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "title": "Hint Pattern"
    },
    "hint_set_by": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "title": "Hint Set By"
    },
    "llm_model": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "title": "Llm Model"
    },
    "llm_provider": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "title": "Llm Provider"
    },
    "mode_requested": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "title": "Mode Requested"
    },
    "mode_resolved": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "title": "Mode Resolved"
    },
    "mode_used": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "title": "Mode Used"
    },
    "prompt_versions": {
      "additionalProperties": {
        "type": "string"
      },
      "title": "Prompt Versions",
      "type": "object"
    },
    "threshold_check": {
      "anyOf": [
        {
          "$ref": "#/components/schemas/ThresholdCheck"
        },
        {
          "type": "null"
        }
      ]
    }
  },
  "title": "AnalysisProvenance",
  "type": "object"
}

Used by endpoints

2 endpoints send or return this schema, on its own or inside another schema.

• POST /api/v1/analyze — Analyze Test Case
• GET /api/v1/analyze/{test_case_id} — Fetch existing AI analysis for a test case (no re-run)

Used by schemas

• AnalysisResponse

More details

• Specification: OpenAPI JSON