What is agent evaluation?
A test of an agent’s whole run: its choices, tool calls, errors, and final result.
05 · Operations · 7 terms
What belongs in operations?
Operations covers tracing, evaluation, state, and human control in production. This section has 7 definitions. Open any term for a plain answer, the source behind it, and the day we checked it. Related links take you to the ideas that explain it or show where it sits in the wider field.
7 shown
What is attestation?
A signed or recorded statement that a required action, check, or condition was completed.
What is evaluation?
A structured test that measures a system against stated criteria and evidence.
What is evaluator?
A person, model, rule, or program that scores an output or run against set tests.
What is observability?
The ability to understand a system from its traces, logs, measures, state, and other clues.
What is observation?
Information an agent gets from its setting before or after it acts.
What is observation space?
The set of signals an agent may get from its setting.