Develop AI Agent Skills from how verification specialists assess an unexpected result using the tested configuration, method suitability, and differences between comparable runs. Help an agent assemble a test-review brief that preserves the original result, compares run conditions, and identifies questions for investigation.

Verification engineers, laboratory specialists, and device quality teams
A verification test produces a result the team did not expect
Carry the test conditions with the test result
A connector fails a mechanical test after an earlier version passed. The device revision, fixture setup, and test software have all changed since that run. The lab needs to preserve the result and explain which comparisons are meaningful before anyone proposes a repeat test.
An unexpected connector result is a starting point for investigation. Make the differences between runs visible without assuming the cause.
Keep the original data, acceptance criteria, observed failure, and protocol deviations together. Separate measured observations from explanations proposed after the result was known.
What did the approved test measure, and where did the observation depart from the expectation?
Compare specimen history, device revision, fixture alignment, calibration status, environmental conditions, and software version. Capture the specialist’s reasoning about which differences could affect the measurement.
Which difference prevents a direct comparison with the earlier passing run?
Document competing explanations and the evidence needed to distinguish them. A repeat result does not erase the original failure. Keep investigation and retest decisions with the authorized test and quality owners.
What would another run resolve that the current evidence cannot?
The test lead and quality reviewer determine the investigation, any authorized retest, and acceptance against the approved protocol.
Review an unexpected result with the laboratory specialist who investigated it. Capture how they separated a possible device issue from method and setup questions. Use that explanation to develop a skill for preparing future verification reviews.
Explore AI Skill EngineeringStart with the original data, the approved protocol, and two runs worth comparing. Include the specialist who can explain the limits of that comparison.