Use this file to discover all available pages before exploring further.
The Google Drive app provides read/write access to a user’s Google Drive through the Drive REST v3 API. It can be used either as a Builder app (automations call Drive instructions directly) or as a remote MCP server consumed by an AI agent. Each tenant configures its own Google OAuth Application; end-users sign in with their own Google account and tokens are stored per (user × tenant). The connector exposes 27 operations covering files (CRUD, copy, export, trash), permissions, comments, replies and revisions.
Files & Folders
List, search, create, update, copy, export and delete files, folders and Google-native documents
Sharing & Permissions
Grant, change and revoke access for users, groups, domains or anyone with the link
Comments & Revisions
Browse and manage comment threads, replies and historical revisions of any file
A Google account with access to the Drive content you want to expose. For Google Workspace drives, the workspace admin may need to allow third-party OAuth apps.
A Google OAuth 2.0 Client of type Web application, created at console.cloud.google.com/apis/credentials. The Authorized redirect URIs must contain the value shown in the OAuth Callback URL field of the installed app instance (auto-populated on install — copy it back into the Google Cloud Console after installation).
An OAuth consent screen configured in the Google Cloud project. While the app is in Testing state, add the end-user’s email to Test users. The full Drive scope (https://www.googleapis.com/auth/drive) is classed restricted and requires CASA verification when the consent screen moves to Production.
Base URL of the Drive API (default https://www.googleapis.com)
API Token (fallback)
Optional static OAuth access token used as a shared fallback for all users of this tenant. Stored as a workspace secret. Most deployments leave this empty and rely on per-user OAuth.
OAuth2 Client ID
Google OAuth Application Client ID. Create an OAuth client of type Web application in the Google Cloud Console. Stored as a workspace secret.
OAuth2 Client Secret
Google OAuth Client Secret, stored as a workspace secret
Space-separated Google API scopes. Default https://www.googleapis.com/auth/drive (full Drive access). Other common values: drive.readonly, drive.file, drive.metadata, drive.appdata
Refresh Token TTL (seconds)
Default 15552000 (180 days, Google’s max)
MCP Endpoint
Auto-populated on install — URL of the MCP endpoint for this instance
MCP API Key
Auto-populated on install — signed key used in the mcp-api-key header. Do not modify
MCP Endpoint and MCP API Key are generated automatically by the onInstall flow. The OAuth credentials (OAuth2 Client ID, OAuth2 Client Secret) must be filled in manually after creating the OAuth client in the Google Cloud Console.
Once the app instance is configured, each end-user authorizes their own Google account through a browser-based consent screen:
1
Trigger the connect flow
From an MCP client (Agent Creator capability or any tool client), call any data tool. If no OAuth session exists for the current user × tenant, the MCP server returns a connector_auth_required payload with a connect_url. Alternatively, call the connect tool explicitly to receive the same payload.
2
Open the connect URL
Open the returned URL in a browser tab where the user is already authenticated to Prisme.ai. The platform redirects to the Google OAuth consent screen.
3
Grant access
The user reviews the requested scopes (default https://www.googleapis.com/auth/drive) and clicks Allow. Google redirects back to the platform’s oauthCallback webhook.
4
Confirmation
The user sees a Connection complete page and can close the tab. The platform has stored an access token plus refresh token as user-scoped secrets — both are tenant-prefixed so they do not leak across app instances.
Each user’s OAuth tokens are scoped per (user × tenant). A user who has authorized in tenant A does not gain access in tenant B — they must run the connect flow again per app instance. Refresh tokens are rotated by Google on each refresh and live up to 180 days when idle.
Every instruction resolves credentials from the workspace configuration. Most list operations accept pageSize (default 100, max 1000) and pageToken for pagination, plus a fields parameter using Google’s partial response syntax to control the returned payload size.
- GoogleDrive.createPermission: fileId: '{{file_id}}' type: user role: commenter emailAddress: alice@example.com sendNotificationEmail: true emailMessage: | Hi Alice — feel free to comment on the Q1 brief. output: permission
The Google Drive app ships with a built-in MCP server. Each app instance gets its own signed mcp-api-key that encodes the workspace ID and a credentials lookup URL — the Google OAuth token is never passed through headers and is resolved server-side from the user’s per-tenant secret.
Agents consume MCP servers directly through Agent Creator capabilities. This is the preferred way to expose Google Drive to an agent.
1
Create or open a workspace
From the Prisme.ai console, create a new workspace (or open the one that will host the connector).
2
Install the Google Drive app
Open the workspace Imports panel, search for Google Drive and install it.
3
Configure the credentials
Open the freshly installed app instance settings and fill in the required fields (see the Usage as App tab for the field-by-field reference). At minimum you need OAuth2 Client ID and OAuth2 Client Secret from your Google Cloud OAuth client.
4
Copy the MCP endpoint and API key
Still on the app instance configuration page, copy the values of MCP Endpoint and MCP API Key — both are generated automatically on install.
5
Open Agent Creator
Switch to Agent Creator and open the agent you want to extend.
6
Add a capability
Add a new capability to the agent:
If a dedicated Google Drive capability exists — select it and paste the MCP API Key into the mcp-api-key field. The server URL is already wired.
Otherwise — select the generic custom_mcp capability, paste the MCP Endpoint into the Server URL field, then open the Headers field and add an mcp-api-key entry whose value is the MCP API Key copied earlier:
{ "mcp-api-key": "your-mcp-api-key"}
7
Save
The agent now has access to every Google Drive tool exposed by the MCP server.
8
Brief the agent in its system prompt
Wiring the capability is not enough — the agent also needs to know the MCP exists and when to reach for it. Add a short paragraph to the agent’s system prompt. Copy-pasteable starter:
You have access to the Google Drive MCP server. Use it whenever the user asks something that maps to Drive data — finding files, browsing folders, reading file metadata, sharing, commenting, or exporting Google Docs/Sheets/Slides. Use `fields` aggressively (e.g. `files(id,name,modifiedTime)`) to keep responses small. Confirm with the user before any destructive action (deleteFile, deletePermission, emptyTrash, deleteComment).
Refine the trigger keywords (resource names, business domains, typical user phrasings) so the agent reliably picks up the right intent in your context.
Use this flow to plug the Google Drive MCP into an AI Knowledge agent that does not yet support the native MCP picker.
1
Install the Google Drive app
Install and configure the app in the same workspace as your agent (see the Usage as App tab). Once configured, mcpEndpoint and mcpApiKey are auto-populated.
2
Copy the MCP credentials
Open the app instance config and copy the values of MCP Endpoint and MCP API Key.
3
Open your AI Knowledge project
Navigate to Advanced > Tools.
4
Add an MCP tool
Click Add and select the MCP tab.
5
Fill in the endpoint
Paste the MCP Endpoint URL copied from the app instance.
6
Add the auth header
In the Headers field, add the signed API key:
{ "mcp-api-key": "your-mcp-api-key"}
7
Save
The agent can now list and call Google Drive tools through the MCP endpoint.
The signed mcp-api-key encodes the workspace ID and the getConfig webhook URL. The MCP server validates the signature using the central app secret, fetches the OAuth client configuration from the installed app, and resolves the end-user’s access token from per-user secrets. Credentials are cached per tenant for 10 minutes.
Initiate the per-user OAuth flow. Returns a connector_auth_required payload with a connect_url. Data tools auto-prompt this when no session is active, so call it directly only when the user explicitly asks to sign in
disconnect
Revoke the current user’s OAuth tokens (RFC 7009) and clear the local secrets
“Not configured” — The app instance has no OAuth client. Fill in OAuth2 Client ID and OAuth2 Client Secret from the Google Cloud Console.“Invalid API key” (MCP) — The mcp-api-key header does not match the central app secret. Reinstall the app instance to regenerate a signed key.“Credentials lookup failed” — The MCP endpoint could not reach the getConfig webhook of the installed app. Verify that the app instance is still installed in the expected workspace.redirect_uri_mismatch at the Google consent screen — The redirect URI registered in your Google OAuth client does not match the one the connector sends. Copy the value of OAuth Callback URL from the app instance configuration and paste it verbatim into the Authorized redirect URIs list of your OAuth client in the Google Cloud Console.Drive API has not been used in project ... or it is disabled — Enable the Drive API at console.cloud.google.com/apis/library/drive.googleapis.com for the project that owns your OAuth client. Propagation takes up to a minute.File not found: <id> on a shared drive — Pass supportsAllDrives: true (and includeItemsFromAllDrives: true on listFiles). Without these flags, the Drive API silently scopes the request to My Drive only.fileNotDownloadable on exportFile — exportFile works only on Google-native files (Docs/Sheets/Slides/Drawings). For uploaded binaries (PDF, DOCX, …), use a file-content endpoint instead.