ss-sdd-fixing-test-failures

Spec-driven development failure fixing applies minimal, narrow-scoped code edits to resolve reported feature test failures without modifying the locked spec or plan. If expectations conflict, it reports the concern rather than rewriting requirements to maintain spec integrity.

Reproduction-driven verification re-runs the tester's exact Given/When/Then reproduction steps for each reported failure. It only reports success when every provided reproduction passes, preventing partial DONE statuses and ensuring the test failure fixing is complete.

Yes, this leaf-agent execution operates without spawning sub-subagents and applies narrow-scoped code changes to fix exactly the listed feature test failures. It escalates only when additional context is required or the right fix is ambiguous.

Preventing partial DONE status requires reproduction-driven verification that re-runs exact reproduction steps for every failure. Status reporting rules ensure DONE is marked only after all provided feature test reproductions pass successfully.

You should not modify the spec or plan when expectations conflict with test failures; instead, report the concern. Spec-driven development integrity requires fixing failures through narrow-scoped code edits while keeping requirements locked.

Yes, reproduction-driven verification requires exact reproduction steps including story, Given/When/Then expectations, and likely locations from a tester subagent. This ensures targeted failure fixing addresses the precise failures without drifting into unrelated refactors.