Use the SDKs to programmatically retrieve logs for debugging, evaluation, or custom analytics pipelines.

​
List Logs

# Assumes quotient client is already initialized (see Initialize the Logger)
logs = quotient.logs.list()
print("Total logs:", len(logs))
for log in logs:
    print(log["id"], log.get("user_query"))

​
Common Fields

Each returned Log object may contain:
  • id (string): Unique identifier for the log.
  • app_name (string): Application that generated the interaction.
  • environment (string): Deployment environment (dev, staging, prod).
  • user_query (string): The prompt that initiated the interaction.
  • model_output (string): Response recorded for the interaction.
  • documents (array): Evidence passed in with the log.
  • message_history (array): Conversation turns, if provided.
  • instructions (array): Instruction set supplied to the model.
  • tags (object): Metadata such as feature, customer, or model version.
  • created_at / updated_at (datetime): Timestamps for auditability.
  • has_hallucination (boolean): Whether hallucination detection flagged the response.
  • evaluations (array): Detailed detection results.

​
Usage Tips

  • Narrow queries with filters (if available) to keep responses manageable.
  • Paginate long-running exports rather than loading the entire history at once.
  • Persist the returned log_id in your own telemetry to cross-reference issues in the Quotient UI.
Next: Async Logging