Smart Connections MCP runs on your own machine — the client starts it, so there is no endpoint to ping. 93 installs a week from npm. Last commit 22 Aug 2026.
Local semantic search over Obsidian vaults via Smart Connections embeddings. Multi-vault, private.
Today is the operative word: we check Smart Connections MCP every 15 minutes and re-read its code on every release. Watch it and you find out the day that stops being true.
This server runs on your own machine — install it with the package manager and the client starts it for you. Package name taken from the official registry entry.
claude mcp add smart-connections-mcp -e 'SMART_VAULT_PATH=<SMART_VAULT_PATH>' -- npx -y smart-connections-mcp
{
"mcpServers": {
"smart-connections-mcp": {
"command": "npx",
"args": [
"-y",
"smart-connections-mcp"
],
"env": {
"SMART_VAULT_PATH": "<SMART_VAULT_PATH>"
}
}
}
}
[mcp_servers.smart-connections-mcp]
command = "npx"
args = ["-y", "smart-connections-mcp"]
env = { SMART_VAULT_PATH = "<SMART_VAULT_PATH>" }
{
"mcpServers": {
"smart-connections-mcp": {
"command": "npx",
"args": [
"-y",
"smart-connections-mcp"
],
"env": {
"SMART_VAULT_PATH": "<SMART_VAULT_PATH>"
}
}
}
}
{
"servers": {
"smart-connections-mcp": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"smart-connections-mcp"
],
"env": {
"SMART_VAULT_PATH": "<SMART_VAULT_PATH>"
}
}
}
}
Put your own value in place of
<SMART_VAULT_PATH> (One or more absolute Obsidian vault paths, comma-separated (alias: SMART_VAULT_PATHS)).
The author’s registry entry marks it as required.
MCP server for Obsidian Smart Connections. Semantic search using your vault's embeddings.
Read-only semantic search MCP server for Obsidian vaults using local Ollama embeddings.
Local RAG search over Obsidian vaults: SQLite hybrid search with local Ollama embeddings.
Local Markdown search, navigation and editing via MCP, with optional vector embeddings.
Search an Obsidian vault by tag, term and wikilink graph. Read-only, deterministic, no embeddings.
Minimal RAG-over-a-corpus MCP retrieval: search_knowledge returns cited chunks. Local embeddings.
Persistent memory for AI agents — semantic + recency search, ONNX embeddings, Docker Compose.
Local hybrid keyword + vector search (SQLite FTS5 + sqlite-vec) over markdown vaults and code.
Answers built from our own checks of this server.