evernote-mcp
A self-hosted Evernote MCP server that lets any MCP client search, read, create, and update notes in your live Evernote account, with notes never passing through third-party servers.
README
evernote-mcp
A self-hosted Evernote MCP server, native to the 2026-07-28 stateless spec.
Point any MCP client at your own machine and it can search, read, create and update notes in your live Evernote account. Your notes never pass through anyone else's server: the only two parties are your client and Evernote.
Built on the official mcp Python SDK 2.0. The 2026-07-28 spec dropped the
initialize handshake and session state, so this server answers a single
stateless POST per call. Clients still on the old spec keep working through
the SDK's fallback path, verified against mcp 1.27.
Tools
| Tool | What it does |
|---|---|
list_notebooks |
Every notebook with guid, name, stack, default flag |
search_notes |
Evernote's own search grammar, passed through unmodified |
read_note |
One note by guid, body converted from ENML to markdown |
create_note |
Title plus markdown body, optional notebook |
update_note |
Replace a note's title, body, or both, by guid |
Deliberately absent in v1: attachments, tags, sync, and anything that deletes. No tool in this server can remove a note.
search_notes takes Evernote query strings verbatim, so
notebook:Recipes intitle:pasta updated:day-7 -tag:archived works exactly as
it does in the Evernote app. An empty query returns the most recently updated
notes.
Setup
Requires Python 3.11+ and uv.
git clone https://github.com/jayzuccarelli/evernote-mcp
cd evernote-mcp
make sync
cp .env.example .env
1. Get an Evernote token
You need a JSON file with at least token and noteStoreUrl:
{
"token": "S=s123:U=...",
"noteStoreUrl": "https://www.evernote.com/shard/s123/notestore"
}
Two ways to produce it:
- Developer token. Evernote issues these from your account settings for personal use. Pair it with the note store URL for your shard.
- OAuth. Run the standard Evernote OAuth flow with your own consumer key.
The access token response carries
edam_webApiUrlPrefix; the note store URL is that prefix withnotestoreappended.
Save the file as ~/.evernote-token.json, or point EVERNOTE_TOKEN_FILE
somewhere else.
2. Set a bearer token
EVERNOTE_MCP_TOKEN is the secret that guards your server. It is unrelated to
the Evernote credential.
uv run python -c "import secrets; print(secrets.token_urlsafe(32))"
Put it in .env. Leave it unset and the server accepts unauthenticated calls,
which is only safe bound to localhost.
3. Run
make run
The server listens on http://127.0.0.1:3402/mcp. Add it to a client:
claude mcp add --transport http evernote http://127.0.0.1:3402/mcp \
--header "Authorization: Bearer $EVERNOTE_MCP_TOKEN"
Exposing it beyond localhost
Put it behind a reverse proxy that terminates TLS. With Tailscale Funnel:
tailscale funnel --bg --set-path /evernote 3402
Then set both of these in .env:
PUBLIC_URL=https://your-host.ts.net/evernote
TRUST_PROXY=true
PUBLIC_URL is what the server advertises in its RFC 9728 protected resource
metadata, so OAuth-aware clients discover the right issuer. TRUST_PROXY
relaxes the DNS-rebinding Host header check, which a proxy forwarding a public
hostname would otherwise trip. Bearer auth still gates every request.
Markdown and ENML
Evernote stores notes as ENML, an XHTML subset. This server converts on the way out and back on the way in, so tools speak markdown.
The conversion is not lossless. Headings, bullets, checkboxes, links, bold,
italic and paragraphs round trip. Tables, images, inline styles and colours do
not. Attachments show up as [attachment: image/png] placeholders on read and
are dropped on write.
update_note replaces the body wholesale. Read first, edit, then write back if
you want to preserve anything.
Development
make check # ruff lint + format check, then import and tool-registry smoke test
make check runs against tests/fake-token.json, so it needs no Evernote
account and touches no network.
License
MIT
Recommended Servers
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.
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.
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.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
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.
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.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
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.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.