This chapter explains how botserver manages knowledge base collections, indexing, caching, semantic search, and conversation memory. The implementation uses vector databases for semantic search and intelligent memory management for context optimization.
Document File Description
README README.md High-level reference for the .gbkb package and its core commands (USE KB, CLEAR KB, USE WEBSITE). KB and Tools kb-and-tools.md Integration patterns for knowledge bases and tool systems. Vector Collections vector-collections.md Definition and management of vector collections, including creation, document addition, and usage in dialogs. Document Indexing indexing.md Process of extracting, chunking, embedding, and storing document vectors in the VectorDB. Retrieval and RAG hybrid-search.md The retrieval pipeline: dense search, keyword fusion, the six modes and their limits. Episodic Memory episodic-memory.md Automatic conversation history management, context compaction, and intelligent summarization. Semantic Caching caching.md Intelligent caching for LLM responses, including semantic similarity matching.
Parameter Default Description
embedding-urlhttp://localhost:8082Embedding service endpoint embedding-modelbge-small-en-v1.5Model for vector embeddings rag-modestandardRetrieval strategy: standard, hybrid, corrective, graph, agentic, multimodal
Parameter Default Description
episodic-memory-enabledtrueEnable/disable episodic memory episodic-memory-threshold4Exchanges before compaction episodic-memory-history2Recent exchanges to keep episodic-memory-modelfastModel for summarization episodic-memory-max-episodes100Max episodes per user episodic-memory-retention-days365Days to keep episodes episodic-memory-auto-summarizetrueAuto-summarize conversations
Parameter Default Description
llm-cachefalseEnable/disable response caching llm-cache-ttl3600Cache time-to-live in seconds llm-cache-semantictrueUse semantic similarity matching llm-cache-threshold0.95Similarity threshold for cache hits
Navigate : Click the file links to read the full documentation for each topic.Reference : Use the parameter tables for quick configuration lookup.Update : When the underlying implementation changes, edit the corresponding markdown files and keep this summary in sync.