API Evangelist enhancements for the i6eal Open AI Data API

overlay: 1.0.0
info:
  title: API Evangelist enhancements for the i6eal Open AI Data API
  version: 1.0.0
extends: openapi/i6eal-open-ai-data-api-openapi.json
x-apievangelist:
  generated: '2026-08-09'
  method: generated
  source: >-
    Derived from live probes of https://i6eal.de on 2026-08-09 plus the provider's
    llms.txt and DCAT catalogue. Records facts that are true of the deployed API but absent
    from the published description. The original document is never mutated.
actions:
  - target: $.info
    description: Record the catalogue rating context and the machine-readable siblings of this spec.
    update:
      x-apievangelist-slug: i6eal-open-ai-data-api
      x-apievangelist-profile: https://apis.io/provider/i6eal-open-ai-data-api/
      x-machine-readable-siblings:
        dcat_catalogue: https://i6eal.de/data/catalog/dcat.jsonld
        llms_txt: https://i6eal.de/llms.txt
        datapackage: https://github.com/i6eal/open-ai-data-catalogue

  - target: $.info
    description: >-
      Make the read-only, unauthenticated posture explicit as structured data rather than
      prose buried in the description.
    update:
      x-access-model:
        authentication: none
        registration_required: false
        rate_limits_published: false
        write_operations: 0
        cors: '*'
      x-agentic-access:
        action-class: read
        consequence: read
        human-in-the-loop: not-required
        audit: not-required
        note: >-
          Every operation is a safe, unauthenticated GET of a public static file. Generated
          by API Evangelist as a governance starting point; not an i6eal claim.

  - target: $.servers[0]
    description: Record the observed delivery characteristics of the production host.
    update:
      x-observed:
        probed: '2026-08-09'
        cdn: Amazon CloudFront over Amazon S3
        tls_version: TLSv1.3
        hsts: max-age=31536000; includeSubDomains
        cache_control: public,max-age=0,must-revalidate
        conditional_requests: ETag + Last-Modified (304 confirmed)
        security_headers:
          - x-content-type-options: nosniff
          - x-frame-options: DENY
          - referrer-policy: strict-origin-when-cross-origin
          - permissions-policy: 'camera=(), microphone=(), geolocation=()'

  - target: $.components.responses.NotFound
    description: >-
      Record what a 404 actually returns on the wire, so generated clients do not attempt to
      parse a JSON error body that is never sent.
    update:
      x-observed-media-type: text/html
      x-observed-note: >-
        Probing a nonexistent distribution returns the site's HTML 404 page, not a
        machine-readable error document. Branch on the status code and Content-Type.

  - target: $.components.schemas.DatasetIndex
    description: Name the shared envelope fields observed across all 16 dataset indexes.
    update:
      x-observed-common-fields:
        - schemaVersion
        - datasetId
        - generatedAt
        - trackingSince
        - license
        - materialFingerprint
        - sourceFingerprint
        - sourcePolicy
        - collectionStatus
        - sourceHealth
        - counts
        - downloads
      x-integrity: >-
        materialFingerprint and sourceFingerprint are SHA-256 digests; compare them across
        fetches to detect a real content change without diffing the payload.

  - target: $.components.schemas.AtomFeed
    description: Record the two path spellings the change feeds actually use.
    update:
      x-observed-paths: [atom.xml, feed.atom]
      x-observed-coverage: 11 of 16 datasets publish a change feed
      x-entry-id-pattern: 'urn:i6eal:<dataset>:event:<YYYY-MM-DD>:<kind>-<hash16>'

  - target: $.tags[?(@.name=='Catalogue')]
    description: >-
      Flag the catalogue tag as the entry point an agent should call first, since it is the
      only operation that reveals the URL set.
    update:
      x-entry-point: true
      x-note: >-
        getOpenDataCatalogue resolves 16 datasets and 82 distributions; every other
        operation's path can be derived from its dcat:accessURL.