{
  "service": "diffbot",
  "displayName": "Diffbot",
  "categories": [
    "Data",
    "AI"
  ],
  "authTypes": [
    "api_key"
  ],
  "auth": [
    {
      "type": "api_key",
      "label": "API Token",
      "placeholder": "DIFFBOT_TOKEN",
      "description": "Diffbot API token sent with the token query parameter. Copy it from the account menu in the Diffbot app.",
      "extraFields": []
    }
  ],
  "homepageUrl": "https://www.diffbot.com",
  "actions": [
    {
      "id": "diffbot.extract_article",
      "service": "diffbot",
      "name": "extract_article",
      "description": "Extract article content and metadata from a public URL using Diffbot Article API.",
      "requiredScopes": [],
      "providerPermissions": [],
      "inputSchema": {
        "type": "object",
        "properties": {
          "url": {
            "type": "string",
            "description": "The public article URL to extract with Diffbot.",
            "format": "uri"
          },
          "fields": {
            "type": "array",
            "items": {
              "type": "string",
              "enum": [
                "links",
                "extlinks",
                "meta",
                "querystring",
                "breadcrumb",
                "quotes"
              ],
              "description": "An optional Diffbot article field to request."
            },
            "description": "Optional Diffbot article fields to include in the response.",
            "minItems": 1
          },
          "timeout": {
            "type": "integer",
            "minimum": 1,
            "description": "The upstream page retrieval timeout in milliseconds."
          },
          "renderDelay": {
            "type": "integer",
            "minimum": 0,
            "description": "The extra delay in milliseconds before extraction."
          },
          "scroll": {
            "type": "string",
            "enum": [
              "fast",
              "slow"
            ],
            "description": "The scrolling mode Diffbot should use to trigger lazy-loaded content."
          },
          "proxy": {
            "type": "string",
            "minLength": 1,
            "description": "The custom proxy IP address Diffbot should use to fetch the target page."
          },
          "proxyAuth": {
            "type": "string",
            "minLength": 1,
            "description": "The proxy authentication string sent with the custom proxy."
          },
          "useProxy": {
            "type": "string",
            "enum": [
              "default",
              "none"
            ],
            "description": "Whether Diffbot should force its default proxy or disable proxy usage."
          },
          "paging": {
            "type": "boolean",
            "description": "Whether Diffbot should automatically concatenate multi-page articles."
          },
          "maxTags": {
            "type": "integer",
            "minimum": 0,
            "description": "The maximum number of auto-generated tags to return."
          },
          "tagConfidence": {
            "type": "number",
            "minimum": 0,
            "maximum": 1,
            "description": "The minimum relevance score required for returned tags."
          },
          "categoryConfidence": {
            "type": "number",
            "minimum": 0,
            "maximum": 1,
            "description": "The minimum relevance score required for returned categories."
          },
          "discussion": {
            "type": "boolean",
            "description": "Whether Diffbot should extract article comments and discussions."
          },
          "naturalLanguage": {
            "type": "array",
            "items": {
              "type": "string",
              "enum": [
                "entities",
                "sentiment",
                "summary",
                "facts",
                "openFacts",
                "records",
                "categories",
                "sentences",
                "language"
              ],
              "description": "A Diffbot Natural Language feature to enable."
            },
            "description": "The Diffbot Natural Language features to enable for the article.",
            "minItems": 1
          },
          "summaryNumSentences": {
            "type": "integer",
            "minimum": 1,
            "description": "The maximum number of summary sentences when summary analysis is enabled."
          }
        },
        "additionalProperties": false,
        "required": [
          "url"
        ],
        "description": "The input payload for extracting one article from a public URL with Diffbot."
      },
      "outputSchema": {
        "type": "object",
        "properties": {
          "request": {
            "anyOf": [
              {
                "type": "object",
                "properties": {
                  "pageUrl": {
                    "type": "string",
                    "description": "The target page URL echoed by Diffbot."
                  },
                  "api": {
                    "type": "string",
                    "description": "The Diffbot API family that handled the request."
                  },
                  "version": {
                    "type": "integer",
                    "description": "The Diffbot API version used for the request."
                  }
                },
                "additionalProperties": true,
                "description": "The top-level request metadata returned by Diffbot."
              },
              {
                "type": "null"
              }
            ]
          },
          "article": {
            "anyOf": [
              {
                "type": "object",
                "properties": {
                  "type": {
                    "type": "string",
                    "description": "The Diffbot object type for the extracted page."
                  },
                  "title": {
                    "type": "string",
                    "description": "The extracted article title."
                  },
                  "pageUrl": {
                    "type": "string",
                    "description": "The canonical page URL extracted from the article."
                  },
                  "text": {
                    "type": "string",
                    "description": "The extracted plain text content of the article."
                  },
                  "html": {
                    "type": "string",
                    "description": "The extracted article HTML content."
                  },
                  "date": {
                    "type": "string",
                    "description": "The article publication date returned by Diffbot."
                  },
                  "humanLanguage": {
                    "type": "string",
                    "description": "The language code returned by Diffbot."
                  },
                  "images": {
                    "type": "array",
                    "items": {
                      "type": "object",
                      "properties": {},
                      "additionalProperties": true,
                      "description": "A Diffbot object with provider-specific fields."
                    },
                    "description": "The images returned by Diffbot for the article."
                  },
                  "tags": {
                    "type": "array",
                    "items": {
                      "type": "object",
                      "properties": {},
                      "additionalProperties": true,
                      "description": "A Diffbot object with provider-specific fields."
                    },
                    "description": "The topic tags returned by Diffbot."
                  },
                  "categories": {
                    "type": "array",
                    "items": {
                      "type": "object",
                      "properties": {},
                      "additionalProperties": true,
                      "description": "A Diffbot object with provider-specific fields."
                    },
                    "description": "The content categories returned by Diffbot."
                  },
                  "naturalLanguage": {
                    "type": "object",
                    "properties": {},
                    "additionalProperties": true,
                    "description": "A Diffbot object with provider-specific fields."
                  }
                },
                "additionalProperties": true,
                "description": "The normalized primary article extracted from the Diffbot article endpoint."
              },
              {
                "type": "null"
              }
            ]
          },
          "humanLanguage": {
            "type": "string",
            "description": "The language code returned at the top level."
          },
          "type": {
            "type": "string",
            "description": "The top-level Diffbot object type."
          },
          "title": {
            "type": "string",
            "description": "The top-level title returned by Diffbot."
          }
        },
        "additionalProperties": false,
        "required": [
          "request",
          "article"
        ],
        "description": "The normalized Diffbot article extraction response."
      }
    }
  ]
}