> ## Documentation Index
> Fetch the complete documentation index at: https://docs.errorbar.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# The certificate before and after adjudication

> Free and uncached: re-reads the stored calibration rows (the judge's verdicts at judging time) against the adjudicated grades — no judge call. `before` is the calibration as measured; `after` is what a re-calibration would read once label_error corrections are applied and ambiguous grades withdrawn. With no adjudications the two are identical. `remaining` lists the open disagreements ranked by error concentration, each with the `label_id` to POST to. Read scope.



## OpenAPI

````yaml /openapi.json get /v1/criteria/{id}/adjudication
openapi: 3.1.0
info:
  title: errorbar Management API
  description: >-
    The management API behind the improvement loop: capture and setup, request
    logs and datasets, grades (labels), judges (criteria), evals and deploy
    gates, fine-tuning and reinforcement learning, dedicated GPU endpoints, and
    model aliases and versions. Authenticated with a workspace API key
    (sk_sovereign_...). The inference API (chat, embeddings, rerank, responses)
    is OpenAI-compatible and documented separately.


    Responses are snake_case, list endpoints on the loop products use the
    {"object": "list", "data": [...]} envelope, and refusals use the same nested
    error shape the gateway emits: {"error": {"message", "type", "code"}}.
    Request bodies on the loop products (logs, labels, criteria, evals,
    datasets, aliases) are snake_case; the training and infrastructure products
    (fine-tuning, GRPO, environment tools, dedicated, model-version adoption)
    validate camelCase bodies, and each schema below says which it is. Endpoints
    that spend money require a key minted by a workspace owner or admin and
    return 403 otherwise.
  version: 1.0.0
servers:
  - url: https://gateway.errorbar.ai
    description: Production
  - url: https://www.errorbar.ai/api
    description: Control plane (also served at this base URL)
security:
  - bearerAuth: []
tags:
  - name: Certified reward
    description: >-
      The certified reward for your own trainer: sessions with judge
      certificates, scoring with a ledger-true kill switch, and the mid-run
      anchor that holds a run when the reward diverges from an independent
      judge.
paths:
  /v1/criteria/{id}/adjudication:
    parameters:
      - name: id
        in: path
        required: true
        schema:
          type: string
    get:
      tags:
        - Judges
      summary: The certificate before and after adjudication
      description: >-
        Free and uncached: re-reads the stored calibration rows (the judge's
        verdicts at judging time) against the adjudicated grades — no judge
        call. `before` is the calibration as measured; `after` is what a
        re-calibration would read once label_error corrections are applied and
        ambiguous grades withdrawn. With no adjudications the two are identical.
        `remaining` lists the open disagreements ranked by error concentration,
        each with the `label_id` to POST to. Read scope.
      operationId: getCriterionAdjudication
      responses:
        '200':
          description: Before, after, and what is left.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/CriterionAdjudication'
        '401':
          $ref: '#/components/responses/Unauthorized'
        '404':
          $ref: '#/components/responses/NotFound'
components:
  schemas:
    CriterionAdjudication:
      type: object
      properties:
        criterion_id:
          type: string
        aligned_at:
          type:
            - string
            - 'null'
        before:
          $ref: '#/components/schemas/CalibrationSnapshot'
        after:
          $ref: '#/components/schemas/CalibrationSnapshot'
        adjudicated:
          type: integer
        by_kind:
          type: object
          properties:
            label_error:
              type: integer
            judge_error:
              type: integer
            ambiguous:
              type: integer
        adjudications:
          type: array
          items:
            $ref: '#/components/schemas/AdjudicationRecord'
        remaining_disagreements:
          type: integer
        remaining:
          type: array
          items:
            $ref: '#/components/schemas/RemainingDisagreement'
          description: >-
            Ranked: the dominant error direction first (one direction carrying
            most of the error is one systematic defect), then the rows the judge
            was most confident about.
        concentration:
          type: array
          items:
            type: object
            properties:
              direction:
                type: string
              count:
                type: integer
              share_of_errors:
                type: number
        certificate_voided:
          type: boolean
          description: >-
            A label_error correction or an ambiguous withdrawal changed the
            measured set: the stored signed certificate is voided until a
            re-calibration; `after` is what that run would read.
    CalibrationSnapshot:
      type: object
      properties:
        'n':
          type: integer
        matrix:
          type: object
          properties:
            tp:
              type: integer
            fp:
              type: integer
            tn:
              type: integer
            fn:
              type: integer
        se:
          type:
            - number
            - 'null'
        se_ci:
          type:
            - array
            - 'null'
          items:
            type: number
        sp:
          type:
            - number
            - 'null'
        sp_ci:
          type:
            - array
            - 'null'
          items:
            type: number
        kappa:
          type:
            - number
            - 'null'
        trust:
          type: string
        holdout_active:
          type: boolean
    AdjudicationRecord:
      type: object
      properties:
        label_id:
          type: string
        request_id:
          type: string
        kind:
          type: string
          enum:
            - label_error
            - judge_error
            - ambiguous
        original_verdict:
          type:
            - string
            - 'null'
        adjudicated_verdict:
          type:
            - string
            - 'null'
        note:
          type:
            - string
            - 'null'
        adjudicated_at:
          type:
            - string
            - 'null'
          format: date-time
    RemainingDisagreement:
      type: object
      properties:
        label_id:
          type: string
        request_id:
          type: string
        judge_verdict:
          type: string
        human_verdict:
          type: string
        critique:
          type:
            - string
            - 'null'
        judge_grade:
          type:
            - number
            - 'null'
        direction:
          type: string
          enum:
            - fp
            - fn
          description: >-
            fp — the judge passed what the human failed; fn — the judge failed
            what the human passed.
        in_report_half:
          type: boolean
    Error:
      type: object
      properties:
        error:
          type: object
          properties:
            message:
              type: string
            type:
              type: string
              description: >-
                invalid_request_error, insufficient_quota, rate_limit_error, or
                api_error.
            code:
              type: string
              description: >-
                Machine-stable cause, e.g. invalid_api_key, not_found,
                insufficient_permissions, precondition_failed.
          required:
            - message
            - type
            - code
      description: >-
        Every refusal — gateway and management API alike — uses this one
        envelope.
  responses:
    Unauthorized:
      description: Missing, malformed, or revoked API key.
      content:
        application/json:
          schema:
            $ref: '#/components/schemas/Error'
          example:
            error:
              message: Invalid API key
              type: invalid_request_error
              code: invalid_api_key
    NotFound:
      description: The resource does not exist in your workspace.
      content:
        application/json:
          schema:
            $ref: '#/components/schemas/Error'
  securitySchemes:
    bearerAuth:
      type: http
      scheme: bearer
      description: >-
        Your workspace API key, e.g. `sk_sovereign_...`, sent as `Authorization:
        Bearer <key>`.

````