AI Demos · Schema

RecordedEvidenceIn

An assessment somebody ELSE already made, transcribed into the archive. `POST /runs/{id}/grade` queues our judge to read the output bytes and form a verdict. The binder has no such option: the verdict it is carrying was reached in June by a harness, or in July by a researcher writing prose, and the bytes that produced it are often gone. Re-judging what we can still see and calling that the historical result would replace the record with our opinion of it. So this endpoint transcribes rather than grades — and `source` is the whole point. `extracted` says *a person or a program stated this and we are recording their words*. `captured` says *this verdict was measured from bytes this run actually produced*, which the June harness did (it compared result sets against a live database) and a researcher's write-up did not. The database enforces the ordering either way: C12 refuses `captured` evidence over an `extracted` run, so the stronger label cannot be attached to a reconstruction even by a caller trying to.

Artificial IntelligenceBenchmarkingMachine LearningDataAnalytics

Properties

Name Type Description
idempotency_key object
scenario_id string
test_case_version_id string
rubric_version_id object the standard this verdict was reached under (C13). For a historical grade that is the rubric version recording the ORIGINAL standard, not today's
actor_id string who judged it — the researcher or the harness, never the binder
verdict string
reasoning string the assessor's own words. Not a summary of them
source string captured | extracted
graded_at string when the assessment was made, not when it was transcribed
score object
measured object the numbers behind the verdict (C3). The gate checks these against the test case's stated expectations
grade_reason string
supersedes object
replicate_of object
observations object
proofs object
View JSON Schema on GitHub

JSON Schema

aidemos-com-recorded-evidence-in-schema.json Raw ↑
{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://artifacts.apis.io/aidemos-com/json-schema/aidemos-com-recorded-evidence-in-schema.json",
  "title": "RecordedEvidenceIn",
  "description": "An assessment somebody ELSE already made, transcribed into the archive.\n\n`POST /runs/{id}/grade` queues our judge to read the output bytes and form a verdict. The binder\nhas no such option: the verdict it is carrying was reached in June by a harness, or in July by a\nresearcher writing prose, and the bytes that produced it are often gone. Re-judging what we can\nstill see and calling that the historical result would replace the record with our opinion of it.\n\nSo this endpoint transcribes rather than grades — and `source` is the whole point. `extracted`\nsays *a person or a program stated this and we are recording their words*. `captured` says\n*this verdict was measured from bytes this run actually produced*, which the June harness did\n(it compared result sets against a live database) and a researcher's write-up did not. The\ndatabase enforces the ordering either way: C12 refuses `captured` evidence over an `extracted`\nrun, so the stronger label cannot be attached to a reconstruction even by a caller trying to.",
  "x-generated": "2026-11-01",
  "x-method": "derived",
  "x-generator": "derive-json-schema.py",
  "x-source": "openapi/aidemos-com-openapi.json#/components/schemas/RecordedEvidenceIn",
  "properties": {
    "idempotency_key": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "title": "Idempotency Key"
    },
    "scenario_id": {
      "type": "string",
      "title": "Scenario Id"
    },
    "test_case_version_id": {
      "type": "string",
      "title": "Test Case Version Id"
    },
    "rubric_version_id": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "title": "Rubric Version Id",
      "description": "the standard this verdict was reached under (C13). For a historical grade that is the rubric version recording the ORIGINAL standard, not today's"
    },
    "actor_id": {
      "type": "string",
      "title": "Actor Id",
      "description": "who judged it — the researcher or the harness, never the binder"
    },
    "verdict": {
      "type": "string",
      "title": "Verdict"
    },
    "reasoning": {
      "type": "string",
      "title": "Reasoning",
      "description": "the assessor's own words. Not a summary of them"
    },
    "source": {
      "type": "string",
      "title": "Source",
      "description": "captured | extracted"
    },
    "graded_at": {
      "type": "string",
      "title": "Graded At",
      "description": "when the assessment was made, not when it was transcribed"
    },
    "score": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "title": "Score"
    },
    "measured": {
      "anyOf": [
        {
          "additionalProperties": true,
          "type": "object"
        },
        {
          "type": "null"
        }
      ],
      "title": "Measured",
      "description": "the numbers behind the verdict (C3). The gate checks these against the test case's stated expectations"
    },
    "grade_reason": {
      "type": "string",
      "title": "Grade Reason",
      "default": "initial"
    },
    "supersedes": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "title": "Supersedes"
    },
    "replicate_of": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "title": "Replicate Of"
    },
    "observations": {
      "anyOf": [
        {
          "items": {
            "$ref": "#/$defs/ObservationIn"
          },
          "type": "array"
        },
        {
          "type": "null"
        }
      ],
      "title": "Observations"
    },
    "proofs": {
      "anyOf": [
        {
          "items": {
            "$ref": "#/$defs/EvidenceProofIn"
          },
          "type": "array"
        },
        {
          "type": "null"
        }
      ],
      "title": "Proofs"
    }
  },
  "type": "object",
  "required": [
    "scenario_id",
    "test_case_version_id",
    "actor_id",
    "verdict",
    "reasoning",
    "source",
    "graded_at"
  ],
  "$defs": {
    "EvidenceProofIn": {
      "properties": {
        "artifact_id": {
          "type": "string",
          "title": "Artifact Id"
        },
        "role": {
          "type": "string",
          "title": "Role",
          "description": "input | output | capture — deliberate subset of artifact_kind: trace/log substantiate a run, not a verdict"
        },
        "span": {
          "anyOf": [
            {
              "additionalProperties": true,
              "type": "object"
            },
            {
              "type": "null"
            }
          ],
          "title": "Span",
          "description": "closed locator contract: exactly one base of pages/start_ms/sheet/json_path/lines/turns + its qualifiers (bbox/end_ms/range)"
        },
        "anchor": {
          "anyOf": [
            {
              "type": "string"
            },
            {
              "type": "null"
            }
          ],
          "title": "Anchor"
        },
        "note": {
          "anyOf": [
            {
              "type": "string"
            },
            {
              "type": "null"
            }
          ],
          "title": "Note"
        },
        "ord": {
          "anyOf": [
            {
              "type": "integer"
            },
            {
              "type": "null"
            }
          ],
          "title": "Ord"
        }
      },
      "type": "object",
      "required": [
        "artifact_id",
        "role"
      ],
      "title": "EvidenceProofIn",
      "description": "#425 — the exact region that substantiates this evidence row. The durable proof is the\nNATIVE artifact + locator; a screenshot crop is `role='capture'`, supporting presentation\nonly, and must already be attached to the run (role='capture') or the database refuses it."
    },
    "ObservationIn": {
      "properties": {
        "text": {
          "type": "string",
          "title": "Text"
        },
        "verdict": {
          "anyOf": [
            {
              "type": "string"
            },
            {
              "type": "null"
            }
          ],
          "title": "Verdict"
        },
        "ord": {
          "anyOf": [
            {
              "type": "integer"
            },
            {
              "type": "null"
            }
          ],
          "title": "Ord"
        }
      },
      "type": "object",
      "required": [
        "text"
      ],
      "title": "ObservationIn"
    }
  }
}

Work with this as data

Every JSON Schema here is available over the APIs.io API and to AI agents over MCP.

MCP server

One button, every client — Claude, Cursor, VS Code and the rest.

https://apis.io/mcp

Tools for schemas

4 MCP tools reach this
  • find_json_schemasBrowse and filter every JSON Schema in the catalog.
  • apis_io_searchSTART HERE — APIs, providers and tags for one query, each with its total.
  • resolveTurn a domain, URL or GitHub org into the provider it belongs to.
  • find_cohortsEvery scored population of providers in the catalog.
All 92 tools →

Call it yourself

curl for this page
This JSON Schema
curl "https://apis.io/api/v1/json-schemas/aidemos-com-recorded-evidence-in"
All schemas
curl "https://apis.io/api/v1/json-schemas?limit=25"

Discovery needs no key. Ratings and market analysis are Pro.

Get an API key

Free tier, no form to fill in. Signing in shares your email address with us — we store it to create your key and to recognise you if you sign in with another provider. See our Privacy Policy and Terms.

A second provider on the same verified email joins the account you already have.