Skip to content

v1.5.1

@anulum anulum tagged this 01 Mar 21:12
HaltEvidence structured recovery on halt, LLM-as-judge hybrid
escalation, cross-encoder reranker, bidirectional AggreFact benchmark,
streaming window sweep ablation, CI regression suite, domain cookbooks,
marketing language cleanup, license FAQ, badge updates.

810 tests passing. Regression suite runs in <1s without GPU.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Assets 2
Loading