Skip to main content

Connect Monologue to ChatGPT, Claude, and other AI apps with MCP

Set up the Monologue MCP server in ChatGPT, Claude, Codex, Cursor, VS Code, Windsurf, Zed, and other AI apps.

N
Written by Naveen Naidu
Updated this week

Monologue's MCP server lets compatible AI apps search and read your Monologue notes after you connect your account.

The server is hosted by Monologue, so you do not need to copy API keys or run anything locally. When your AI app connects, it opens a Monologue sign-in and consent screen. After you approve it, the app can use Monologue tools to find notes, list recent notes, and open note details.

The MCP server address is:

https://api.monologue.to/mcp

What you get

Once connected, your AI app can help with questions like:

  • "Find my notes about the hiring plan."

  • "What did I say in recent notes about the launch?"

  • "Search my notes from last week for customer feedback."

  • "Open the note about the design review and summarize it."

The first version of Monologue MCP is read-only. It can read notes from the Monologue account you authorize, but it cannot create, edit, delete, or share notes.

Before you start

Make sure you can sign in to the Monologue account that contains the notes you want to use.

You should also use an AI app that supports MCP. Some apps support remote MCP servers directly. Others use `mcp-remote`, a small compatibility package that lets apps connect to hosted MCP servers through a local command.

ChatGPT

Monologue may not appear in ChatGPT's connector directory yet. For now, add it as a custom MCP connector.

  1. Open ChatGPT.

  2. Go to Settings.

  3. Open Apps & Connectors or Connectors.

  4. If needed, open Advanced and turn on Developer mode.

  5. Add a custom MCP connector.

  6. Name it `Monologue`.

  7. Enter the remote MCP server URL:

https://api.monologue.to/mcp

Then save the connector and complete the Monologue login and consent flow when ChatGPT opens it.

After it is connected, start a new chat and choose Monologue from the connectors or tools menu. You can also ask ChatGPT to use Monologue in your prompt, for example: "Search my Monologue notes for the design review."

If you use ChatGPT Business, Enterprise, or Edu, your workspace may require an admin or owner to add and approve custom connectors before members can use them.

Claude

Monologue may not appear in Claude's connector directory yet. For now, add it as a custom connector.

  1. Open Claude.

  2. Go to Settings.

  3. Open Connectors.

  4. Click Add connector or Add custom connector.

  5. Name it `Monologue`.

  6. Enter the connector URL:

https://api.monologue.to/mcp

Then click Add, click Connect, and complete the Monologue login and consent flow.

After it is connected, enable Monologue from Claude's connectors menu in the chat where you want to use it.

If you use Claude Team or Enterprise, an owner may need to add Monologue from organization connector settings before members can connect their own Monologue accounts.

Codex

Run this in Terminal:

codex mcp add monologue --url https://api.monologue.to/mcp

Codex will open the Monologue login and consent flow. Sign in, approve access, and return to Codex.

If this is your first remote MCP server in Codex, add this to `~/.codex/config.toml` first:

[features]
experimental_use_rmcp_client = true

You can also configure it manually in `~/.codex/config.toml`:

[features]
experimental_use_rmcp_client = true

[mcp_servers.monologue]
url = "https://api.monologue.to/mcp"

Then run:

codex mcp login monologue

Claude Code

Run this in Terminal:

claude mcp add --transport http monologue https://api.monologue.to/mcp

Then open a Claude Code session and run:

/mcp

Choose Monologue and follow the login flow.

Cursor

Open Cursor's MCP settings and add a custom server.

Use this configuration:

{
"mcpServers": {
"monologue": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://api.monologue.to/mcp"]
}
}
}

Save the configuration, start the Monologue MCP server from Cursor's MCP tools, and complete the Monologue login flow when prompted.

Visual Studio Code

Open the command palette and search for `MCP: Add Server`.

  1. Choose Command (stdio).

  2. Enter:

npx mcp-remote https://api.monologue.to/mcp

Then name the server `Monologue`, start it from `MCP: List Servers`, and complete the Monologue login flow when prompted.

You can also use this configuration:

{
"mcpServers": {
"monologue": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://api.monologue.to/mcp"]
}
}
}

Windsurf

Open Windsurf settings, go to Cascade, then open MCP servers.

Add a custom server with:

{
"mcpServers": {
"monologue": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://api.monologue.to/mcp"]
}
}
}

Start the server and approve Monologue access when the browser login flow opens.

Zed

Open Zed settings and add:

{
"context_servers": {
"monologue": {
"source": "custom",
"command": "npx",
"args": ["-y", "mcp-remote", "https://api.monologue.to/mcp"],
"env": {}
}
}
}

Save the settings, start the context server, and complete the Monologue login flow.

Other MCP clients

If your app supports remote Streamable HTTP MCP servers, use:

https://api.monologue.to/mcp

If your app only supports local stdio MCP servers, use:

Privacy and access

Monologue MCP uses OAuth, so the AI app asks you to sign in and approve access instead of asking for a personal API token.

The current permission is `notes:read`. That means the connected app can read the Monologue notes available in your account. Note details can include titles, summaries, timestamps, and transcript text when requested by the AI app.

Only connect Monologue to AI apps you trust. Once an AI app reads a note through MCP, that app handles the note content according to its own privacy and data policies.

Troubleshooting

Opening the MCP URL in a browser shows an authentication error

That is expected. The MCP URL is meant to be opened by an MCP client, not directly in a normal browser tab.

The app asks for an API key

Monologue MCP does not use personal API keys. Use the hosted MCP URL and complete the browser login flow instead.

The login flow gets stuck or fails

Try removing saved MCP auth data and connecting again:

rm -rf ~/.mcp-auth

Then restart your AI app and reconnect Monologue.

The `mcp-remote` command fails

Make sure Node.js is installed and up to date, then try starting the MCP server again.

Did this answer your question?