When an agent has been given real permissions -- to read files, call tools, edit records -- the question "what did it actually do, and when" needs a real answer, not a best guess reconstructed after the fact. An audit log is the plain, append-only record of every action an agent took, in the order it took them.
Write build_audit_log(actions) that turns a list of (timestamp, actor, description) tuples into formatted lines of the form "[timestamp] actor: description", preserving order.
No results yet
Run your code to check it against the examples, or submit to run every test.