Two agents (impl_a: airouter, impl_b: best-complexity model) implement
in parallel against pre-written tests using TDD composites. An audit
node compares both outputs and emits a JSON verdict (pass/more_work_needed,
winner: a/b/neither, gaps). Harnesses are param-overridable for
provider flexibility. F-WE-PDIV note in description for future
unique_providers enforcement.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
