5
Verification
Ensuring the agent's output is correct, complete, and safe.
4 patterns
5.1
Incremental Verification
Verify each unit of work as it is produced, not the whole batch at the end, so errors surface while they are cheap to fix and trivially attributable.
Beginner4 min read
5.2
Adversarial Review
Have a separate agent or a fresh context attack the work, because the agent that produced it shares the blind spots that produced the bug.
Intermediate4 min read
5.3
Regression Guard
Pin the existing behavior with characterization tests before the agent touches the code, so "it still works" is a check the machine runs, not a hope you hold.
Intermediate4 min read
5.4
Diff Review
Read every changed line before accepting. The agent's summary is its intent; the diff is what it actually did — and they are not the same artifact.
Beginner5 min read