v0.6.0

Minor Changes

• 95fdd7b Thanks @coopbri! - Wire advanced RAG pipeline from agent-core: LLM query condensing, BM25+embedding hybrid search with reciprocal rank fusion, and optional cross-encoder reranking for knowledge retrieval

• 00c6830 Thanks @coopbri! - Add per-turn knowledge selection so relevant knowledge chunks are injected based on each user message instead of only at init, and migrate shared knowledge types to agent-core