SLATENO
Log in

MCP · CONNECT YOUR ASSISTANT

Your assistant.
Your workspace.

Turn a conversation into work you can open and edit. Connect an AI assistant you already use to your decks, documents and notes in SLATENO.

Set up your connection

For the desktop app. Keep it open while your assistant works.

FROM A REQUEST TO A REAL FILE

Your AI assistantA client on your computer
MCP
SLATENO desktop
DecksDocsNotes
The connection is local. Your assistant may use an online AI provider to process content.

A COMMON LANGUAGE

What is MCP?

The Model Context Protocol lets AI applications work with tools and information from another app. In SLATENO, it gives a connected assistant a defined set of actions for your workspace.

You bring the assistant and its AI service. SLATENO provides the workspace tools. Connecting one does not turn on built-in AI generation or include a subscription to an AI provider.

MAKE THE CONNECTION

Start here. Then ask away.

You need the desktop app, a signed-in account, and an MCP-compatible client running on the same computer.

  1. 01

    Open the assistant settings

    In the desktop app, go to Account → Preferences → Connect an AI assistant. Expand Set up an assistant (MCP).

  2. 02

    Switch on the connection

    Enable Allow AI agents to connect (MCP). Wait for the connection status to show that it is listening. Turn on Read-only if you want to start with searching and summaries.

    Your access token is created here. Keep it private, just as you would a password.

  3. 03

    Add your client

    Under Connect a client, choose your assistant and copy the configuration shown in the app. It contains the correct app path or local address for your setup. Follow the matching instructions below.

    Choose your client
  4. 04

    Try a simple request

    Ask your assistant: “List my decks in SLATENO. Don’t change anything.” Check the result and look under Recent activity in the app to confirm a tool call arrived.

    When you are ready to create or edit, switch off Read-only. The current connection supports direct edits, so review the result in SLATENO.

CHOOSE YOUR CLIENT

Use the assistant you know.

Copy the configuration from the desktop app rather than copying someone else’s token or guessing a port.

ClaudeDesktop and Claude Code

Claude Desktop

  1. In SLATENO, expand Claude under Connect a client and copy the Claude Desktop configuration.
  2. In Claude Desktop, open Settings → Developer → Edit Config.
  3. Merge the generated entry into mcpServers, keeping your existing servers. Save and restart Claude Desktop.

This uses the app’s local command bridge. Keep SLATENO open.

Official local-server setup

Claude Code

Copy the Claude Code command from the same section in SLATENO, run it in a terminal, then use /mcp in Claude Code to check the connection. The command contains your token, so keep it private.

Claude Code’s MCP guide
ChatGPTDesktop with local MCP support
  1. In SLATENO, expand ChatGPT under Connect a client.
  2. In the ChatGPT desktop app, open Settings → MCP servers → Add server and choose STDIO.
  3. Use the executable path from command and add --mcp-stdio as the argument. Save, then select Restart.
  4. Use /mcp to check the connection, then try a request in a local task.

You can also merge the generated TOML entry into your personal ~/.codex/config.toml. These steps apply to desktop clients with local MCP settings. ChatGPT web does not read this local configuration; its remote connection needs separate setup.

Official ChatGPT MCP guide
CursorLocal MCP settings
  1. Copy the Cursor configuration from SLATENO.
  2. Add its server entry through Cursor’s MCP settings or your personal ~/.cursor/mcp.json file. Merge it with any existing mcpServers.
  3. Enable the server in Cursor, confirm its tools are available, and try a request in the agent chat.

Use personal configuration for a token-bearing entry. A project’s .cursor/mcp.json may be shared with others.

Cursor’s MCP guide
PerplexityMac app with local MCP access
  1. Open Settings → Connectors in the Perplexity Mac app. Install the official PerplexityXPC helper when prompted.
  2. Choose Add Connector → Simple and name the connector SLATENO.
  3. Copy the Perplexity Mac command from SLATENO and paste it into the command field.
  4. Save, wait for Running, then enable the connector under Sources in a new conversation and try a request.

Keep SLATENO open. Local MCP must be available in your Perplexity account and Mac app; these instructions do not apply to the website.

Perplexity’s local MCP guide
KimiKimi Code on your computer
  1. Copy the Kimi Code configuration from SLATENO.
  2. Merge the server entry into your user-level ~/.kimi-code/mcp.json. You can also use /mcp-config to add a stdio server with the provided command and argument.
  3. Start a new Kimi Code session, then use /mcp to check the server status.
  4. Keep SLATENO open and ask Kimi Code to list your decks.

This setup uses Kimi Code as the local MCP client. It does not connect the Kimi website to your computer.

Kimi Code’s MCP guide
Other MCP clientsFor a custom local setup

Use a client that supports a local command server or Streamable HTTP. The desktop app provides a command bridge, a Python agent example and the local HTTP connection details.

Command clients run the generated executable path with --mcp-stdio. HTTP clients use the displayed 127.0.0.1 address, the /mcp path and the bearer token. Use a compatible MCP client to handle initialization and tool calls.

A service running only in the cloud cannot directly reach this local address. There is no hosted MCP endpoint provided by this guide.

MCP transport reference

GIVE IT SOMETHING USEFUL TO DO

Small requests. Useful results.

Start with a test project. Be specific about the file and the changes you want.

READ-ONLY

Get your bearings

“List my decks in SLATENO and summarize the most recently updated one. Don’t edit it.”
WRITING ENABLED

Turn an outline into slides

“Create a five-slide deck called ‘Project kickoff’ from this outline. Keep each slide to three short points.”

Add your outline to the request.

WRITING ENABLED

Keep the useful bits

“Find my document called ‘Project brief’, summarize its decisions, and save them in a new note.”

Available tools cover decks, documents and notes. Design, Tasks, Calendar and Boards are not currently exposed through this connection. There is no delete or export tool.

KNOW WHAT YOU’RE CONNECTING

Your connection.
Your controls.

Read the privacy details

Access covers the workspace

A connected client can read supported decks, documents and notes across your workspace. There is no per-file permission picker. Read-only blocks writes, but it still allows reading and searching.

Changes can happen directly

With writing enabled, tools can create content, add slides and replace document text. MCP does not provide a separate approval preview for every edit. Keep backups and check the result.

Local connection, separate provider

SLATENO exposes workspace tools on your computer. Your assistant may send the content it reads to its AI provider, under that provider’s settings and terms. These tools do not grant general filesystem or shell access.

Stop or reconnect at any time

Disable the connection to stop access, or regenerate the token and update clients that use it. The command bridge reads the current token locally. Closing SLATENO stops the server. Administrators can disable MCP or require read-only mode.

IF SOMETHING GETS STUCK

Check these first.

Can’t connect?

Keep the desktop app open and signed in. Confirm MCP is enabled and listening, then copy fresh connection details. Restart or reconnect your client after changing its configuration.

Unauthorized or refused?

For an authorization error, update the client’s token. If reading works but edits are refused, check Read-only and any administrator restriction.

No assistant settings?

Look in the desktop app, not the website. If an administrator has switched MCP off, the app shows that status. Check that you are using the current desktop app.

Ready to connect?

Start in the desktop app. Need a hand? The Help page opens our feedback channel.

Get the desktop app Get help