What problem does it solve? When a change spans several extensions, a later test failure is ambiguous: did the change break something, or was it already broken? This Skill establishes a recorded pre-change baseline by running each extension's own test harness before any source edit, so later failures can be checked against durable evidence instead of memory. ## Core Features & Use Cases - Per-extension harness execution: Runs each named extension's test/run.sh separately so failures stay traceable to the specific extension that produced them. - Verbatim result recording: Captures exact pass/fail counts per harness, or the raw exit code and full output when no explicit count is printed. - Durable baseline persistence: Writes the baseline into the task's artifact, report, or log so later steps can compare against it. - Use Case: Before refactoring code shared by three extensions, run all three harnesses, record their exact results, and later prove that a reported failure was pre-existing rather than a newly introduced regression. ## Quick Start Run each named extension's test harness individually, record the exact results, and save them as a pre-change baseline before editing any source files.
