Custom Connectors

1. In Abacus.AI, open Agent Settings and select MCP Server Config from the sidebar.
2. Add a JSON entry for Conductor that includes the server URL and an Authorization header carrying your API token:

{
  "conductor": {
    "url": "https://mcp-universal.conductor.com/mcp/v3",
    "headers": {
      "Authorization": "Bearer YOUR_CONDUCTOR_API_TOKEN"
    }
  }
}

3. Save the configuration. Abacus queries the server and lists Conductor’s available tools.

​

Get your client secret from Conductor

​

Prerequisites

• Set the Disable custom MCP server connector for Gemini Enterprise organization policy constraint to Off (Not enforced). This requires an Organization Policy Administrator.
• Add all three of Conductor’s hosts to the allowedEgressFqdns parameter in your egress domain policy: mcp-universal.conductor.com, auth.conductor.com, and prod-conductor.auth.us-east-1.amazoncognito.com. Miss either of the last two and the OAuth exchange fails even though the MCP server itself is reachable.
• Add custom_mcp to the allowed data sources, if project enforcement is enabled or the project is protected by VPC Service Controls.
• Grant the Discovery Engine Editor (roles/discoveryengine.editor) role to whoever creates the data store.

​

Create the data store

1. In the Google Cloud console, go to Gemini Enterprise.
2. In the navigation menu, click Data stores, then click Create data store.
3. On the Select a data source page, enter Custom MCP Server in the Search sources field, then click Add MCP server on the card that appears.
4. In the Authentication settings section, select OAuth 2.0 and complete the form:
   • MCP Server URL: https://mcp-universal.conductor.com/mcp/v3
   • Authorization URL: https://auth.conductor.com/oauth2/authorize
   • Authorization URL Parameters: leave empty. Gemini Enterprise appends the standard OAuth parameters itself.
   • Token URL: https://prod-conductor.auth.us-east-1.amazoncognito.com/oauth2/token
   • Client ID: 2ff2jkrpif0dmb7meqc4tc0o4d
   • Enable PKCE Support: on. Conductor’s authorization server supports PKCE with the S256 challenge method—turn this off only if the token exchange errors.
   • Client Secret: [Enter the client secret you received from Conductor]
   • Scopes: openid profile email
5. Click Verify Auth, then click Continue.
6. Select Multi-region as the location, enter a name in Your data connector name (for example, Conductor), and click Create.

​

Enable Conductor’s tools

1. Open your new Custom MCP data store.
2. Click Actions > Reload custom actions to fetch the list of available tools.
3. Select the tools you want, then click Enable actions. You can enable up to 100 at a time.
4. Connect the data store to the app or agent that should use it. See Google’s guide to connecting a data store to an app for the steps.
5. Authorize the connection: open the app, select Manage your data, click Enable actions next to the Conductor data store, and sign in with your Conductor credentials.

1. Open your librechat.yaml configuration file.
2. Add Conductor as a streamable-http server, passing your API token in the Authorization header. Set requiresOAuth: false so LibreChat uses your header instead of attempting an OAuth flow:

mcpServers:
  conductor:
    type: streamable-http
    url: https://mcp-universal.conductor.com/mcp/v3
    requiresOAuth: false
    headers:
      Authorization: "Bearer YOUR_CONDUCTOR_API_TOKEN"

3. Restart LibreChat so it picks up the new server, then select Conductor in the MCP menu.

1. In Lovable, open Connectors and go to Chat connectors.
2. Click New MCP server.
3. Enter a Server name (for example, Conductor) and set the Server URL to https://mcp-universal.conductor.com/mcp/v3.
4. Under Authentication, choose Bearer token or API key and paste your Conductor API token.
5. Click Add server. Conductor now appears in your list of chat connectors.

1. Add an MCP Client Tool node to your workflow and connect it to your AI Agent node.
2. Set the SSE Endpoint (server URL) to https://mcp-universal.conductor.com/mcp/v3.
3. Set Authentication to Bearer Auth, then create a credential whose token is your Conductor API token. If you prefer to send the token yourself, choose Header Auth and add an Authorization header with the value Bearer YOUR_CONDUCTOR_API_TOKEN.
4. Save and execute the workflow. The agent can now call Conductor’s tools.

1. A workspace admin enables custom connections in Settings > Notion AI > AI connectors > Enable Custom MCP servers.
2. Open your Custom Agent’s Settings and go to Tools & Access.
3. Click Add connection > Custom MCP server.
4. Enter the MCP server URL https://mcp-universal.conductor.com/mcp/v3 and a display name (for example, Conductor).
5. When prompted for authentication, choose header-based auth and add an Authorization header with the value Bearer YOUR_CONDUCTOR_API_TOKEN.
6. Click Save. Conductor’s tools become available to the agent.

1. In Replit, open the MCP Servers settings pane and click Add MCP server.
2. Paste the server URL https://mcp-universal.conductor.com/mcp/v3.
3. Open Advanced settings and add a custom header:
   • Key: Authorization
   • Value: Bearer YOUR_CONDUCTOR_API_TOKEN
4. Click Test & save. Once the connection succeeds, Conductor appears under MCP Servers and its tools are available to Agent.