.svg)
What is Data Logging?
Data logging is the systematic recording of events, inputs, outputs, and metrics over time so systems can be audited, debugged, and improved.
For ML and LLM applications, logs should capture enough context to reproduce failures—prompts, tool calls, model versions, and outcomes—while respecting retention and privacy policies.
Related reading on Giskard
Log enough to audit agent behavior — Pair request/response logging with red-team evidence so you can replay failures and prove whether a policy was violated. Learn more: Practical guide on AI security and LLM vulnerabilities.
Authoritative reference: Wikipedia: Data logger
