Skip to main content
Emitted when an identification is finalized. Match data.identification_id to your browser result and deduplicate deliveries by event_id. The example uses schema_version: "2026-10-13". Use the received event_type and schema_version when parsing a delivery. See Webhook reference for the envelope, signature and retry behavior.

Example payload

The shortened example below shows VPN and Anti-Detect Browser: 30 + 60 = 90. Unshown flags and traffic fields are omitted for readability.

Identity and context fields

Local IP is an address exposed by the browser network check, not a promise to return a private LAN address.

Signals and flags

  • risk_score: authoritative 0–100 result, or 999 for Banned IP.
  • signals: named score contributions. Contributions can be consolidated or corrected; do not reconstruct the final score by summing them.
  • detection_flags: 20 boolean decisions in the 2026-10-13 schema, including zero-weight detections. The example omits most false keys for readability; a full delivery includes them. Risk signals lists every key and its nominal weight.
For this schema, os_mismatch covers either operating-system mismatch check.

Identification score and current identity risk

data.risk_score describes this identification. The analytics dashboard separately shows current Visitor risk, User risk, Device risk and IP risk based on each identifier’s identification history. These four current categories are not fields in the scored webhook. History can return them in identity_risk on an identification card, or in risk.band on the corresponding identity card. Do not interpret hre.*.level as one of those categories: it is the confidence of an account event. See identity risk.

Bot ownership

ShieldLabs identifies the company behind a recognized search bot, AI bot or AI browser. In data, search_bot_owner, ai_bot_owner or ai_browser_owner contains one company name when the corresponding signal is set and the provider is known. These fields do not identify the program or the person controlling it, and do not change the Risk Score. The webhook does not send the string Unknown; the dashboard may display Unknown when a name is absent. Optional client_identity contains additional client attribution. See Bot provider.

Client identity

When supplied, client_identity separates recognized claims from verified attribution. Its presence does not make every claim verified. The full path is data.client_identity. Integrations must handle absent or null nested fields and empty arrays; missing metadata does not mean a human visitor or successful verification. Read claims for the presented name and verified for what the evidence establishes. Assessments can be verified, unverified, unknown or conflict. A provider attribution can be verified while a specific agent-name claim remains unverified. data.client_identity.assessments is an array of individual results, rather than one overall verification status. candidate_profile_id, when present, refers to a claim’s profile_id; an entry without it describes a general assessment. An empty array supplies no assessment verdict. See Identity assessment. The client identity glossary explains the nested profile, attribution, evidence, version and timestamp fields. See AI assistant identity, browser-platform verification and the supported bot directory.

High-Risk Events

data.hre contains three separate account evaluations: Multi-accounting uses its own group-change event. It is not a fourth result inside this object. Event confidence and the identification’s numeric Risk Score are separate measurements.

Result fields

Each account evaluation has the following fields: The integration glossary defines status, level, reason, cluster_id and the account-sharing counts in plain language.

Status and confidence

For example, reason: "no_history" or "skipped" accompanies an evaluation that was not completed. Anonymous checks use not_applicable with reason: "anonymous_user". These are result-context examples, not a closed list of reason values. A null cluster_id means no published cluster reference was supplied. It does not override a non-null detection level, and it is not an identifier to synthesize from the account or device.

Account-sharing counts

Account sharing can include these additional fields: An illustrative account_sharing result:
This example reports a detection and a lower-bound count. It does not make four devices a universal confidence threshold. A current account risk category is also separate from the event’s confidence and the score of this check. See High-Risk Events.