Skip to main content

Verifying your setup

To confirm that the setup is correct, go to your AI client and ask: "What Salesforce org am I connected to with Cirra AI?".

You should see something like:

Successful connection verification

You should see a Cirra ai init tool call.

If for some reason it does not appear, you can be more explicit and prompt the LLM with init Cirra AI.

Make sure the correct tools are enabled​

If you have issues, first make sure the correct tools are enabled.

  • For Claude ensure the Cirra AI connector is enabled

    Claude connector enabled

  • For ChatGPT you should be in developer mode and have Cirra AI enabled

    ChatGPT tools enabled
Troubleshooting

If the tools are connected but you receive a message like this:

Connection error message

This typically indicates that the connection to the Cirra AI MCP Server was not successful.

Check the connection as explained just above, and if that does not work then redo the steps under Connecting to your AI client.

For other issues, you should receive more specific error messages.

Tool Permissions​

AI clients will typically ask your permission the first time before using any tool provided by the Cirra AI MCP server:

Tool permission dialog

Click Always allow to prevent repeated popups for the same function.

Granting tool permission in bulk​

To proactively grant permissions and disable these dialogs:

  • For Claude go to https://claude.ai/settings/connectors and click Configure for the Cirra AI connector.

    You should then see a dialog to change the default tool permissions

    Tool permission dialog

    You can click 'Expand' in each section to change the permissions on a more granular level.

  • ChatGPT as of now does not have a way to grant permissions by default

  • For other AI clients, follow the instructions provided by your vendor