Vault Search FAQs

Question 1

Does Vault Search require an API key for its features?

Accepted Answer

No, Vault Search uses local ONNX embeddings by default, meaning no API key is needed. It also offers support for OpenAI-compatible APIs like Ollama, LiteLLM, OpenAI, and OpenRouter for those who prefer API-based embeddings.

Question 2

What is Vault Search and how does it work?

Accepted Answer

Vault Search is a container-native tool providing hybrid search for Obsidian vaults. It combines BM25 keyword matching, fuzzy title search, and vector semantic search, merging results using Reciprocal Rank Fusion (RRF) for highly relevant outcomes.

Question 3

How does Vault Search improve search accuracy in Obsidian?

Accepted Answer

It significantly enhances accuracy by using hybrid search with RRF, which intelligently combines keyword, fuzzy, and semantic results. Additionally, it supports optional cross-encoder re-ranking and knowledge graph traversal for deeper relevance.

Question 4

Can I use Vault Search with different embedding models?

Accepted Answer

Yes, Vault Search supports multi-model embeddings. You can switch between different models at any time, and it maintains per-model vector tables within a single portable SQLite database.

Question 5

Is Vault Search easy to integrate into existing workflows?

Accepted Answer

Absolutely. It offers a comprehensive command-line interface (CLI) for direct use, supports dual MCP transport (stdio for local clients like Claude Desktop/Cursor, HTTP/SSE for Docker), and is container-native for easy deployment.

Vault Search

Vault Search

Key Features

Use Cases

Key Features

Use Cases