feat: enhance RAG functionality with query rewriting, result reranking, and additional metadata in responses

README.md

@@ -134,8 +134,22 @@ history-api/
 
     OPEN_ROUTER_API=
     OPEN_ROUTER_MODEL=
+    OPEN_ROUTER_FALLBACK_MODEL=qwen/qwen3-30b-a3b-instruct-2507
     OPEN_ROUTER_EMBEDDING_MODEL=
 
+    RAG_QUERY_REWRITE_ENABLED=true
+    RAG_REWRITE_HISTORY_TURNS=3
+    RAG_RETRIEVAL_CANDIDATES=30
+    RAG_CONTEXT_TOP_N=8
+    RAG_GENERATION_MAX_RETRIES=2
+    RAG_GENERATION_RETRY_DELAY_MS=500
+    RAG_RERANK_ENABLED=true
+    RAG_RERANK_MODEL=cohere/rerank-4-pro
+    RAG_RERANK_FALLBACK_MODEL=cohere/rerank-4-fast
+    RAG_RERANK_TIMEOUT_SECONDS=10
+    RAG_RERANK_MAX_RETRIES=2
+    RAG_RERANK_RETRY_DELAY_MS=250
+
     GOONG_API_KEY_MAP=
     GOONG_API_KEY_REQ=
 
@@ -214,4 +228,4 @@ This project is licensed under the MIT License - see the [LICENSE](LICENSE) file
 - [pgvector](https://github.com/pgvector/pgvector) - Vector similarity search for PostgreSQL
 - [LangChain Go](https://github.com/tmc/langchaingo) - Framework for LLM applications
 - [Swagger UI](https://swagger.io/tools/swagger-ui/) - Interactive API documentation
-- [Rustfs](https://github.com/rustfs/rustfs) - High performance S3 compatible object storage
+- [Rustfs](https://github.com/rustfs/rustfs) - High performance S3 compatible object storage