trilium-mcp

trilium-mcp

An MCP server that lets AI agents read and write a self-hosted TriliumNext knowledge base over its ETAPI, enabling note capture, structured table maintenance, and search.

Category
Visit Server

README

trilium-mcp

An MCP server that lets AI agents (Claude Desktop, Claude Code, any MCP-compatible client) read and write a self-hosted TriliumNext knowledge base over its ETAPI.

Single static Go binary. No runtime dependencies. Talks to your local Trilium over HTTP(S) and to the client over stdio.

Why

TriliumNext is a strong personal KB: tree-of-notes with attributes (labels, relations) that double as table columns / board lanes / calendar events. This MCP exposes the right slice of ETAPI so an agent can:

  • Capture stuff into your notes (reading lists, decisions, research dumps).
  • Maintain structured "tables" by creating notes-as-rows under a parent and tagging them with labels-as-columns.
  • Search your existing KB and feed snippets back into a conversation.

It is intentionally minimal: ten tools, ~600 lines of Go, zero clever abstractions.

Tools

Tool Purpose
create_note Create a note (optionally under a parent, with labels in one shot).
batch_create_notes Create many notes in one call — saves per-call schema overhead during restructuring.
get_note Fetch note metadata; optionally include body content.
get_note_subtree Recursively fetch a note + descendants up to N levels as a nested tree — replaces N+1 get_note calls.
update_note Partial update: include only the fields you want to change; omitted fields stay as-is.
append_content Append text to the body with a configurable separator.
delete_note Delete a note and its subtree.
batch_delete_notes Delete many notes; partial failures don't stop the rest.
move_note Re-parent a note in two ETAPI calls (vs the old read-recreate-delete dance).
clone_note Add the note under an additional parent — Trilium-native multi-parent links.
delete_branch Remove one parent-child link without deleting the note (un-clone).
search_notes Full-power Trilium search (#label, ~relation, note.title %= "regex", ancestor scoping, etc.).
add_label Attach a label (#key=value) — acts as a "column" in collection views.
add_relation Attach a relation (~name → noteId) — like a foreign key between notes.
remove_attribute Remove a label or relation by its attribute id.
list_attributes List all labels and relations on a note.

Quick start

1. Run TriliumNext

If you don't already have one:

# docker-compose.yml
services:
  trilium:
    image: triliumnext/notes:latest
    ports:
      - "8092:8080"
    volumes:
      - ./data:/home/node/trilium-data
docker compose up -d

Open http://localhost:8092/, finish the setup wizard, then Options → ETAPI → Create new ETAPI token. Copy the token (shown only once).

2. Install trilium-mcp

Pre-built binary (recommended) — grab the right archive from Releases.

From source with Go 1.23+:

go install github.com/OVDEN13/trilium-mcp@latest

With Docker (no Go on host):

git clone https://github.com/OVDEN13/trilium-mcp && cd trilium-mcp
docker build -t trilium-mcp .

3. Configure

Copy .env.example to .env next to the binary:

TRILIUM_URL=http://localhost:8092
TRILIUM_TOKEN=your-etapi-token-here
# Optional:
# TRILIUM_HTTP_TIMEOUT_SECONDS=30

Or pass the same as real environment variables — the server reads either.

4. Register with your MCP client

Claude Code (CLI):

claude mcp add --scope user trilium /path/to/trilium-mcp \
  --env TRILIUM_URL=http://localhost:8092 \
  --env TRILIUM_TOKEN=your-token

Claude Desktop — add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or the equivalent on your OS:

{
  "mcpServers": {
    "trilium": {
      "command": "/absolute/path/to/trilium-mcp",
      "env": {
        "TRILIUM_URL": "http://localhost:8092",
        "TRILIUM_TOKEN": "your-token"
      }
    }
  }
}

Restart the client. The ten tools should show up as trilium__*.

Usage patterns

"Database" of notes (the killer feature)

Trilium's collection views (Table / Board / Calendar) render any note's children based on shared labels. So a "table" is just a parent note + child notes + a consistent label schema:

Books (parent)
├── "Atomic Habits"   #status=read    #rating=9   #author=Clear
├── "Antifragile"     #status=read    #rating=8   #author=Taleb
└── "Деньги"          #status=reading             #author=Жонсон

An agent populates it like this:

// 1. Create the row
create_note({
  parent_note_id: "<id of Books>",
  title: "Atomic Habits",
  labels: { "status": "read", "rating": "9", "author": "Clear" }
})

// 2. Query rows later
search_notes({ query: "#status=read #rating>=8", ancestor_note_id: "<id of Books>" })

Flip the parent's view to Table (or Board by status, or Calendar by a date label) in the Trilium UI and you have a database without ever leaving notes.

Append-only log

append_content({ note_id: "<journal id>", content: "Decided to ship v0.2 on Monday." })

append_content is non-destructive — handy for daily journals, decision logs, ideation dumps.

Trilium search cheat sheet

  • #tag — note has label tag.
  • #status=active — label equals.
  • #rating>=8 — numeric comparison.
  • ~author.title *= "Clear" — follow a relation, match relation target's title.
  • note.title %= "^Re:" — regex on title.
  • note.content *= "kubernetes" — substring in body.
  • #status=active OR #status=pending — boolean.
  • Combine with ancestor_note_id to scope to a subtree.

Full reference: Trilium search docs.

Environment variables

Var Default Notes
TRILIUM_URL required Base URL of your Trilium instance, e.g. http://localhost:8092. The /etapi path is added automatically, but a trailing /etapi is tolerated and stripped (so http://localhost:8092/etapi also works). Accepts multiple URLs separated by commas — the server tries them in order and falls back to the next one on transport errors (DNS/connection/timeout). HTTP errors like 404 are returned immediately without retry. Example: http://192.168.0.10:8092,https://memo.example.com (fast LAN first, public fallback).
TRILIUM_TOKEN required ETAPI token from Trilium settings
TRILIUM_HTTP_TIMEOUT_SECONDS 30 Per-request timeout
TRILIUM_MCP_LOG info off / info / debug. Logs are written to stderr (stdout is reserved for the MCP JSON-RPC stream). info shows one line per tool call with name + duration + ok/error. debug also shows the request arguments and a truncated preview of the response.

Building from source

git clone https://github.com/OVDEN13/trilium-mcp
cd trilium-mcp
go build -ldflags="-s -w" -o trilium-mcp .

Cross-compile (e.g. for macOS from Linux):

GOOS=darwin GOARCH=arm64 CGO_ENABLED=0 go build -o trilium-mcp-darwin-arm64 .

Security notes

  • The server reads TRILIUM_TOKEN from env. Treat it like a password — anyone with it can read and write your entire KB. Keep .env out of git (it is in .gitignore).
  • The binary speaks only to your configured Trilium URL. It does not phone home, log to disk, or open any listening ports.
  • HTTPS works automatically (the binary ships with system CAs when run from the host; the Docker image includes ca-certificates).

Contributing

PRs welcome. Useful directions:

  • Stream large note bodies instead of buffering.
  • move_note / clone_note tools.
  • Bulk operations (add_label_to_many).
  • ETAPI v2 features as TriliumNext adds them.
  • Tests against an ephemeral TriliumNext container.

For substantive changes, please open an issue first to discuss the shape.

License

MIT.

trilium-mcp is an independent project; it is not endorsed by or affiliated with the TriliumNext project. TriliumNext itself is AGPL-3.0; this MCP server talks to it only over its public ETAPI.

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured
E2B

E2B

Using MCP to run code via e2b.

Official
Featured