Bug analysis

Overview

Bug analysis correlates validation failures with issue tracking systems (Jira, Linear, Azure DevOps). Cluster related failures by stack signature, route, and agent category to prioritize root-cause work.

Test Health signals feed bug analysis with flakiness scores and regression confidence.

Step-by-step procedure

Separate observation from the proposed cause

Record the expected behavior, actual behavior, source revision, environment, inputs, and available evidence. A stack trace or AI-generated explanation is a starting hypothesis, not proof of root cause.

Reduce the failure to the smallest safe reproduction. Compare a failing and passing condition, and check recent application, dependency, configuration, and test changes. Keep observations separate from assumptions so another engineer can assess the reasoning.

Verify a fix against the original failure

A fix should address the reproduced failure and include checks for adjacent behavior that could regress. Review changes to tests as carefully as changes to application code; changing the expected result can conceal a defect.

Record the reviewed change, verification conditions, outcome, and remaining uncertainty. If the failure cannot be reproduced, describe the limit of the evidence and the monitoring or further investigation needed.

Related documentation