Review Feedback

After a validation invocation completes, reviewers can attach structured feedback to each item's validation findings. This guide covers listing, creating, and updating violations on a completed item.

Overview

Each validation finding (ValidateItemViolation) on a completed item is either:

• system - generated automatically by the validation pipeline. You cannot create these via the API, but you can review them.
• user - created by your application on behalf of a human reviewer. These can be created and edited.

The REST API base URL is https://brand-intelligence.adobe.com. Violations sit at the item level under the skills/validate resource hierarchy:

GET    /skills/validate/{invocation_id}/items/{item_id} → item detail incl. violations
POST   /skills/validate/{invocation_id}/items/{item_id}/violations
PATCH  /skills/validate/{invocation_id}/items/{item_id}/violations/{violation_id}

data-variant=info

data-slots=text

List violations

Violations aren't listed on their own endpoint - fetch the item detail, which includes all violations (both system and user) under result.violations.

curl --request GET \
  --url "https://brand-intelligence.adobe.com/skills/validate/$INVOCATION_ID/items/$ITEM_ID" \
  --header "Authorization: Bearer $ACCESS_TOKEN"

Response (200 OK):

{
  "itemId": "a1b2c3d4-0000-0000-0000-000000000002",
  "status": "completed",
  "mediaType": "image/png",
  "itemName": "banner-02",
  "thumbnailUrl": "https://example.com/signed/banner-02-thumb.png",
  "renditionUrl": "https://example.com/signed/banner-02.png",
  "result": {
    "summaryText": "The asset violates 1 brand guideline.",
    "violations": [
      {
        "violationId": "123e4567-e89b-12d3-a456-426614174002",
        "violationSource": "system",
        "violationSummary": "Logo clear space violation detected.",
        "assetAttributions": [],
        "corpusAttributions": [],
        "reviewStatus": null,
        "rejectionReason": null,
        "rejectionDetails": null,
        "userName": null,
        "recordVersion": 1
      }
    ],
    "passes": [],
    "rejections": []
  }
}

Each violation may also include assetAttributions and corpusAttributions linking back to the specific regions in the asset and the guideline chunks that informed the finding. reviewStatus is null until a reviewer accepts or rejects the violation. The result also includes passes and rejections lists with the same finding schema. See the API Reference for the full ValidateItemViolation schema.

Add a violation

Create a user-authored violation on a completed item.

curl --request POST \
  --url "https://brand-intelligence.adobe.com/skills/validate/$INVOCATION_ID/items/$ITEM_ID/violations" \
  --header "Authorization: Bearer $ACCESS_TOKEN" \
  --header 'Content-Type: application/json' \
  --data '{
    "violationSummary": "Logo clear space is insufficient on mobile breakpoint."
  }'

Do not include userName; it is not part of the create request schema.

Response (201 Created): the created violation object with violationSource: user and a recordVersion.

Update a violation

Apply a feedback action to an existing violation using PATCH. The action field selects one of four request shapes: acceptViolation, rejectViolation, resetViolation, or editViolation.

Every update requires an If-Match header containing the violation's current version as a quoted string. Read recordVersion from the item detail or the created violation:

export VIOLATION_ID=123e4567-e89b-12d3-a456-426614174002
export RECORD_VERSION=1

All update examples below use that version. Each successful update returns 204 No Content, not a JSON body. Use --include to see the response headers, then update RECORD_VERSION to the version in ETag (for example, ETag: "2" means export RECORD_VERSION=2) before another update. If another reviewer has changed the finding, fetch the current item detail and reconcile the change before retrying with its version.

Accept a violation

curl --include --request PATCH \
  --url "https://brand-intelligence.adobe.com/skills/validate/$INVOCATION_ID/items/$ITEM_ID/violations/$VIOLATION_ID" \
  --header "Authorization: Bearer $ACCESS_TOKEN" \
  --header "If-Match: \"$RECORD_VERSION\"" \
  --header 'Content-Type: application/json' \
  --data '{
    "action": "acceptViolation"
  }'

Reject a violation

curl --include --request PATCH \
  --url "https://brand-intelligence.adobe.com/skills/validate/$INVOCATION_ID/items/$ITEM_ID/violations/$VIOLATION_ID" \
  --header "Authorization: Bearer $ACCESS_TOKEN" \
  --header "If-Match: \"$RECORD_VERSION\"" \
  --header 'Content-Type: application/json' \
  --data '{
    "action": "rejectViolation",
    "rejectionReason": "incorrectAssessment",
    "rejectionDetails": "The logo spacing meets the updated brand guidelines v2.1."
  }'

Reset a violation

Clear a violation's feedback back to pending (un-accept / un-reject):

curl --include --request PATCH \
  --url "https://brand-intelligence.adobe.com/skills/validate/$INVOCATION_ID/items/$ITEM_ID/violations/$VIOLATION_ID" \
  --header "Authorization: Bearer $ACCESS_TOKEN" \
  --header "If-Match: \"$RECORD_VERSION\"" \
  --header 'Content-Type: application/json' \
  --data '{
    "action": "resetViolation"
  }'

Edit a violation

Update the summary text of a violation:

curl --include --request PATCH \
  --url "https://brand-intelligence.adobe.com/skills/validate/$INVOCATION_ID/items/$ITEM_ID/violations/$VIOLATION_ID" \
  --header "Authorization: Bearer $ACCESS_TOKEN" \
  --header "If-Match: \"$RECORD_VERSION\"" \
  --header 'Content-Type: application/json' \
  --data '{
    "action": "editViolation",
    "violationSummary": "Updated finding description."
  }'

What's next

• See the full schema for violations in the API Reference.
• Return to Core Concepts for an overview of the Invocation → Item → Violation hierarchy.