Run Security Audit From Trace
Security
Run Security Audit From Trace
Run security audit on a trace file.
Automatically extracts user input / assistant response pairs from the trace and evaluates them for safety violations (S1-S23).
Request Body:
{
"project_name": "my-project",
"trace": { ... trace data ... },
"categories": ["S1", "S2", "S3"],
"batch_name": "Optional batch name"
}
Trace Format Support:
- Messages with role: user/assistant
- Messages with type: human/ai
- Spans with LLM call input/output
- Execution tree with final response
POST
Run Security Audit From Trace
Authorizations
Valiqor API key. Pass as Bearer token: Authorization: Bearer vq-xxxxxxxxxxxxxxxxxxxx
Body
application/json
Request to run security audit from a trace
Project name (will be created if doesn't exist)
Trace data containing messages/spans
Optional batch name
Categories to check (e.g., ['S1', 'S2'])
Optional configuration
Optional OpenAI key override
Response
Successful Response