AgentDB Vector Search

This implements semantic search using AgentDB's vector database, which claims 150x to 12,500x faster operations than traditional solutions with sub-100 microsecond queries. You get HNSW indexing, binary quantization for 32x memory reduction, and built-in support for OpenAI embeddings or custom models. The CLI tools are genuinely useful for quick database init, querying, and import/export without writing code. Use it for RAG systems where you need to retrieve relevant context before generating answers, or any search that needs to understand meaning rather than just match keywords. The hybrid search combining vector similarity with metadata filters is the standout feature here, letting you constrain semantic results by date, category, or other structured data.