Skip to content

Usage

Python client

from onetrust_api.auth import get_client

client = get_client()  # reads ONETRUST_* environment variables

# Every OpenAPI operation is a method on the composite Api client.
incident = client.create_incident_using_post(body={"name": "Data breach"})
print(incident.status_code, incident.data)

# List endpoints paginate automatically (offset or cursor):
records = client.get_data_subjects_v4(page=0, size=200)

Every call returns a Response wrapper exposing .data (decoded JSON) plus .status_code and .headers.

MCP server

Each domain is a single action-routed tool. Pass an action and a params_json payload (path, query, and body fields all go in params_json):

// tool: onetrust_incidents
{
  "action": "create_incident_using_post",
  "params_json": "{\"body\": {\"name\": \"Data breach\"}}"
}
// tool: onetrust_dsar
{ "action": "get_request_queues", "params_json": "{\"page\": 0, \"size\": 50}" }

The custom_api tool (onetrust_api_request) issues arbitrary requests for any endpoint not yet wrapped:

{ "method": "GET", "endpoint": "/api/incident/v1/incidents",
  "params_json": "{\"params\": {\"page\": 0}}" }

A2A agent

onetrust-agent --provider openai --model-id gpt-4o --api-key sk-...

The agent auto-discovers the MCP tools from mcp_config.json; with this multiple domains the scion-sdk graph router engages automatically to keep the active tool set focused per turn.