Report types reference¶

This page lists every report type the Report Studio can generate, with the id to use from code and the data each one needs. For how to customize, brand, sign off, share and schedule reports, see Reports.

All report types generate as PDF, Word (DOCX) or HTML.

Report types¶

Id	Name	Scope	Needs before it can be generated
`quarterly_board_pack`	Quarterly Board Pack	Organization	At least one active model
`iso42001_statement_of_applicability`	ISO/IEC 42001 Statement of Applicability	Framework: ISO/IEC 42001	ISO/IEC 42001 applicability decisions
`eu_ai_act_annex_iv_technical_documentation`	EU AI Act Annex IV Technical Documentation	One AI system	An EU AI Act risk classification of high for that system
`eu_ai_act_conformity_checklist`	EU AI Act Conformity Checklist	Framework: EU AI Act	EU AI Act applicability decisions
`gdpr_dpia_automated_decision_record`	GDPR Art. 35 DPIA / Art. 22 Automated-Decision Record	Framework: GDPR	Article 30 processing records or automated-decision requests
`ccpa_admt_risk_assessment`	CCPA ADMT Risk Assessment	Framework: CCPA	Applicability decisions for the CCPA ADMT controls
`nist_ai_rmf_profile`	NIST AI RMF Profile	Framework: NIST AI RMF	NIST AI RMF applicability decisions
`model_card`	Model Card	One AI system	An authored model card for that system
`fairness_audit`	Fairness Audit	One AI system	At least one bias audit for that system
`drift_report`	Drift Report	One AI system	At least one drift report for that system
`evidence_pack`	Evidence Pack	Organization	At least one active model
`auditor_traceability_matrix`	Auditor Traceability Matrix	Organization	Control applicability decisions for at least one framework

Applicability decisions are made under Compliance & Evidence → Controls (see Compliance & Evidence). When a report type's needs aren't met, it reports Needs data with the specific reason. It never produces a partial document filled with placeholders.

Generate from code¶

Python SDKREST

types = client.reports.list()
for t in types:
    print(t["id"], t["name"])

html = client.reports.preview(report_type_id="drift_report", model_id=model_id)
export = client.reports.generate(report_type_id="drift_report", model_id=model_id, format="pdf")

# List report types, with availability for your organization
curl https://api.whiteboxxai.com/api/v1/report-types \
  -H "Authorization: Bearer $WHITEBOX_API_KEY"

# Preview as HTML
curl -X POST "https://api.whiteboxxai.com/api/v1/report-types/drift_report/preview?model_id=<model-id>"       -H "Authorization: Bearer $WHITEBOX_API_KEY"

# Generate; the response is an export record with an id
curl -X POST "https://api.whiteboxxai.com/api/v1/report-types/drift_report/generate?format=pdf&model_id=<model-id>"       -H "Authorization: Bearer $WHITEBOX_API_KEY"

format (pdf, docx or html), model_id, brand_profile_id and template_id are query parameters. To override a template's writing style for one run, send the settings as the JSON body.

The generated report is an export: download it from GET /api/v1/export/exports/{export_id}/download, and share or sign it off as described in Reports.

Pass model_id only for report types scoped to one AI system. Organization- and framework-scoped types ignore it.

Generate from an AI assistant¶

An AI assistant generates reports through the Python SDK, using the calls above. The MCP server doesn't have report-generation tools; over MCP, an assistant can create and read drift reports and read the evidence a report draws on. Have the assistant take report type ids from client.reports.list() rather than guessing, and give you the download link and generation date. The compliance-evidence skill covers this.