Production behaviors that were never covered by the eval suite — discovered through user incidents.
A practical guide to untested agent behaviors — what it is, what causes it, how to stop it before it ships harm, and how to catch it when prevention fails.
What it actually looks like in production
- Agent handled an input pattern eval didn't include
- Eval missed combinations of input + retrieval state
- New user segment encountered behaviors not in test set
Why it happens
- Eval set frozen too early
- Test inputs don't reflect production distribution
- Bug reports not fed back into eval
How to prevent it (vendor-neutral)
1. Continuously grow eval set from production incidents
2. Production sampling into eval set
3. Adversarial inputs in eval
4. Coverage tracking per intent / scenario
How Prefactor helps detect and prevent it
Prefactor sits at the agent runtime and contributes specifically:
- Runtime guardrails that flag or block matching patterns before they land
- Continuous eval suites that catch quality regressions on every change
- Tamper-evident logs of every incident and response action
- Per-agent anomaly alerts on the signals listed below
Detection — what to monitor
- Quality on eval set high but incidents in production
- Bug reports clustering in untested areas
Response — what to do when it happens
Immediate (minutes): confirm the incident from the trace; pause the affected agent if active harm possible; hotfix the trigger.
Short-term (hours): add the failure case to the eval suite; patch the root cause; redeploy with regression validation.
Medium-term (days): root cause analysis; tighten guardrails or controls; document the incident for post-mortem and audit.
FAQ
Can untested agent behaviors be eliminated entirely? Usually no — reduce frequency and severity dramatically, and contain blast radius. Aim for low, detected, and contained.
How often should we test for this? Continuously, with every change. Every reported incident becomes a test case.
Can Prefactor detect this in real time? Yes for many variants — guardrails run in-line with sub-second latency.
Related
See Prefactor in action
[Get started free →] [Book a demo →]