Obsidian MCP Server
by Minhao-Zhang
This Obsidian plugin runs a local MCP (Model Context Protocol) server, allowing external applications to interact with your Obsidian vault through a standardized interface. It enables semantic search and other interactions with your notes.
Last updated: N/A
What is Obsidian MCP Server?
The Obsidian MCP Server is an Obsidian plugin that runs a local Model Context Protocol (MCP) server. This allows external applications, such as AI assistants or scripts, to interact with your Obsidian vault through a standardized interface, primarily for semantic search and content manipulation.
How to use Obsidian MCP Server?
- Configure the plugin settings in Obsidian, including the embedding model details. 2. Index your vault using the 'Re-index Vault (MCP Server)' command. 3. Start the MCP server. 4. Connect your MCP client to the server endpoint. 5. Utilize the available MCP tools (e.g.,
simple_vector_search,list_files,read_file) from your client to interact with your vault.
Key features of Obsidian MCP Server
Runs a local SSE-based MCP server.
Indexes vault content into an Orama vector database for semantic search.
Uses a configurable OpenAI-compatible embedding model.
Provides commands to start/stop the server and re-index the vault.
Offers a settings tab for configuration of server, embedding model, and indexing parameters.
Exposes tools like
simple_vector_search,count_entries,list_files,read_file,create_file, andedit_file.
Use cases of Obsidian MCP Server
Integrating AI assistants with your Obsidian vault for question answering and note generation.
Automating tasks such as file creation, editing, and content retrieval using scripts.
Building custom tools that leverage semantic search capabilities within your vault.
Connecting Obsidian to other applications that support the MCP protocol.
FAQ from Obsidian MCP Server
What is MCP?
What is MCP?
MCP stands for Model Context Protocol. It's a standardized interface for interacting with models and data sources.
What embedding models are supported?
What embedding models are supported?
The plugin supports any OpenAI-compatible embedding model, including OpenAI and local Ollama models.
How do I re-index my vault?
How do I re-index my vault?
Use the 'Re-index Vault (MCP Server)' command from the Obsidian command palette.
What if the indexing process fails?
What if the indexing process fails?
If your vault contains a lot of notes, the indexing process might fail due to the database size limit. Try reducing the number of notes or using a different vector database that supports sharding.
Where can I find the server endpoint?
Where can I find the server endpoint?
The server endpoint is displayed in the plugin settings (e.g., http://localhost:8080/sse).