An audit log records who did what, when and with what result: each model call, tool call, approval and denial. For AI agents it answers the questions that matter afterwards, such as which agent changed a record and who approved it, and it is often required for compliance. A log is only useful if it cannot be changed quietly, which is why tamper-evident designs chain each entry to the one before.
Logs also feed monitoring: detection rules can flag injection attempts, unusual tool use or data leaving the system, and a response plan says what happens next. These guides cover tamper-evident audit logs in Python, audit trails for authenticated MCP servers, detecting attacks from logs, and incident response in a production security checklist.
The library lists all guides with filters for topic, level, category and reading time.