AI agent testing glossary

Definitions for the vocabulary around testing AI agents, including the terms Mirrors uses in its own product and pricing. Each page opens with the definition itself, then covers how the thing works and how it differs from whatever it gets confused with.

Last updated 2 August 2026

Agent regression test
A replay of a recorded session against a change, checking that behavior a team already relies on still holds.
Agent staging environment
A non-production copy of the systems an agent calls, so its actions land on fabricated data.
Replay-minute
One minute of sandbox time an agent spends replaying a session or running an eval. The unit Mirrors meters.
Seeded environment
Fabricated data with the shape of the real thing: same schema and relationships, invented values.
Tool-call drift
The gap between the tool calls in a recording and the calls a replay makes after a change.