flaky-build-investigation

Diagnose intermittent build, test, packaging, or CI pipeline outcomes by comparing equivalent runs and correlating logs, configuration, code, environment, ordering, and timing. Use when a build or test sometimes fails, a rerun passes, behavior depends on runner or execution order, or the user asks why the build is flaky. Default to a bounded read-only investigation. Do not use for consistently failing builds, unrelated runtime defects, or release and deployment execution.