v1/classify

Detect personal or sensitive information in a payload without modifying it. Use the body fields below to control which elements are reported. To redact rather than detect, use v1/scrub.

Classify

post

Detect personal or sensitive information within any given payload

Authorizations

x-api-keystringRequired

Body

contentstringOptional

Data you would like to classify

Default: Hi, my name is John Doe, you can reach me on [e-mail]

filtersstring[]Optional

List of data elements to detect in the payload. If set to None, will detect all data elements Teleskope supports

Responses

200

200

application/json

400

400

application/json

post/v1/classify

POST /v1/classify HTTP/1.1
Host: api.scanner.demo-001.teleskope.ai
x-api-key: YOUR_API_KEY
Content-Type: application/json
Accept: */*
Content-Length: 96

{
  "content": "Hi, my name is John Doe, you can reach me on [e-mail]",
  "filters": [
    "text"
  ]
}

{
  "classifications": [
    {
      "index": 0,
      "class": "given",
      "category": [
        "PROFILE"
      ],
      "startChar": 8,
      "charSize": 4,
      "startByte": 8,
      "byteSize": 4,
      "version": 1
    },
    {
      "index": 1,
      "class": "family",
      "category": [
        "PROFILE"
      ],
      "startChar": 13,
      "charSize": 3,
      "startByte": 13,
      "byteSize": 3,
      "version": 1
    },
    {
      "index": 2,
      "class": "email",
      "category": [
        "PROFILE"
      ],
      "startChar": 20,
      "charSize": 20,
      "startByte": 20,
      "byteSize": 20,
      "version": 1
    }
  ],
  "ml_classifier_used": true,
  "scanner_version": "stable-2026-08-03"
}

Body options

Field	Type	Description
content	string	Text to classify. Required.
filters	string[]	Allow-list of element labels to report. Omit to report every supported element.
excludeFilters	string[]	Deny-list of element labels to skip. Takes precedence over filters.
sensitivityLevels	string[]	Restrict results to these tiers: low, medium, high, critical. Elements with no assigned tier are always included.
useML	boolean	Enable ML-based detection in addition to rules. Defaults to true.
treatAsJSON	boolean	Treat content as a JSON string: classify only inside string values. Use for JSON payloads. The root must be an object; wrap a top-level array as {"items": [...]}.

Response fields

Each entry in classifications describes one detected element.

Field	Type	Description
index	integer	Position of this result within classifications.
class	string	Element label, for example email. Use this value with filters and excludeFilters.
category	string[]	Categories the element belongs to.
startChar	integer	Character offset of the match within content.
charSize	integer	Length of the match in characters.
startByte	integer	Byte offset of the match within content.
byteSize	integer	Length of the match in bytes.
version	integer	Taxonomy version used for this result.

Use startChar and charSize in languages with character-indexed strings, and startByte and byteSize when slicing raw bytes. The two differ whenever the payload contains multi-byte characters.

The response also carries scanner_version, ml_classifier_used, and timing.

Earlier revisions of this page showed data_element, star_char, and char_size. The API has never emitted those names. Code written against them parses a valid response and finds nothing. Read class, startChar, and charSize.

Element labels

filters and excludeFilters accept element labels from the Entity Taxonomy.

Names and email addresses are medium tier. Setting sensitivityLevels to ["high", "critical"] alone excludes them; use ["medium", "high", "critical"] to cover everyday personal data.

Related

• v1/classifyCollection — classify multiple values in one request.

• v1/scrub — redact instead of detect.

• Redaction Methodologies — how each redaction style transforms a detected value.