Connect an AI agent to your notes

Let a compatible AI tool read your Monologue notes and manage public share links through MCP.

Monologue can connect supported AI agents to your Voice Notes through the Model Context Protocol (MCP). This lets the agent find and read notes you authorize without copying each transcript by hand.

The connection uses Monologue’s hosted MCP service and an authorization screen in your browser. You do not need to create a separate API key. A connected agent can search and read available note content. If you also grant notes:write, it can turn a public share link on or off for notes you own. It cannot edit or delete your notes through MCP.

Set up a connection

On Mac, open Settings → Notes.
Find Connect an AI Agent.
Choose your AI tool and use the current setup command shown by Monologue.
Approve the connection in your browser when prompted.
Return to the AI tool and ask it to list a recent note.

Use Monologue’s in-app setup command rather than an older command copied from a guide—the available clients and authentication flow can change.

Some AI tools can connect to a remote MCP service directly. Others use a local compatibility helper. Follow the current option shown for your tool in Monologue instead of reusing another client’s command.

Available tools

Tool	Needs	What it does
`list_recent_notes`	`notes:read`	Lists your recent notes and notes shared with you.
`search_notes`	`notes:read`	Searches your notes and notes shared with you.
`get_note`	`notes:read`	Reads one note, optionally with its transcript.
`set_note_share`	`notes:read notes:write`	Turns the public share link for a note you own on or off.
`get_shared_note`	`notes:read`	Reads a public shared note by its share ID, including speakers and owner name.

What an agent can read

A connected agent may receive note titles, summaries, timestamps, and transcripts for notes available to your Monologue account. Only connect an AI tool when you trust both the tool and the people who can access its conversations.

list_recent_notes, search_notes, and get_note also return each note’s share fields:

Field	Meaning
`share_status`	`enabled` for a live public link, `revoked` for a turned-off link, or `null`.
`share_id`	The public share ID when the link is enabled; otherwise `null`.
`share_url`	The public link when sharing is enabled; otherwise `null`.

Reading a note never turns sharing on. A note ID is not a public link: to read a shared note, get_shared_note needs its share ID. Missing, revoked, and private shares return a not-found error.

The agent still decides how to use the results it reads. Review its answer before sharing or acting on it.

set_note_share works only on notes you own, and only when the connection was approved with notes:write.

Turning sharing on publishes the note to anyone with the returned link. Asking again keeps the same link.
Turning sharing off revokes the link and clears its share ID and URL. Anyone who opened the note only through that link loses access.
Turning sharing on again creates a new link. The old link stays invalid.

Ask the agent to confirm before it publishes a note, and treat a returned link like any other share link from Monologue.

Connections approved before sharing was available have read-only access. Refreshing that connection cannot add notes:write; you need a new approval.

Make sure the AI tool requests notes:read notes:write offline_access. Keep offline_access so the tool can stay signed in.
Disconnect the Monologue MCP connection in the AI tool.
Reconnect and approve the new permissions in your browser.
Ask the agent to share a note again.

You do not need to remove or re-register the app in Monologue. Tools that request only notes:read keep working as before; they just cannot change sharing.

Disconnect

Remove the Monologue MCP connection from your AI tool. You can also revoke the connection from Settings → Notes in Monologue. Revoking access prevents that authorization from reading notes or changing share links again.

If you reconnect later, use the current setup flow and approve a new authorization. Do not delete a shared MCP authentication folder or run a broad cleanup command: that can sign other services out as well.

If the connection stops working

Confirm that the Monologue connection is still enabled in the AI tool.
Disconnect only the Monologue connection.
Revoke its authorization under Settings → Notes.
Run the current setup flow again and approve it in your browser.

If the tool opens a browser but does not return to the connection flow, retry once with your default browser and confirm that pop-ups or redirects were not blocked.