Skip to main content

Connect Muse to DataDoe

Connect Meta Muse to DataDoe to ask about your Amazon sales, advertising, and inventory. You sign in to DataDoe during setup.

Before you start

  • You have a DataDoe account with your Amazon accounts connected.
  • You can sign in to Muse.
  • Your role allows MCP OAuth. See Manage roles and access.

Muse connects to DataDoe as a custom connector. DataDoe sign-in uses a public OAuth client. There is no client secret.

If your Organization uses a different server address, copy it from Integrations > MCP (opens in a new tab) and use it everywhere this guide shows https://mcp.datadoe.com/mcp/v1. See What is DataDoe MCP?.

Step 1: Add the connector

  1. In Muse, open Settings, then Connectors, and select Add custom connector. You can also ask Muse in chat to add a custom connector.
  2. Enter DataDoe as the name.
  3. Paste this server address.
https://mcp.datadoe.com/mcp/v1
  1. Select remote (Streamable HTTP) as the transport.
  2. Select OAuth as the authentication method.

If Muse asks for more OAuth values, enter them as follows.

  • Leave the client secret empty.
  • Set token authentication to none.
  • Set the PKCE method to S256.
  • Set the resource to the same server address.
  • Set the scope to mcp:read mcp:write offline_access.

offline_access lets Muse refresh the connection. Sign-in requires the resource, and that resource must be the server address you entered.

Step 2: Enter a DataDoe client ID

Some Muse versions ask for a client ID before you can add the connector. Register a public client with DataDoe, then paste the client_id from the response. That value is the client ID.

Register the client from a terminal. This creates a client with no secret:

curl -X POST https://auth.api.datadoe.com/auth/register \
  -H 'content-type: application/json' \
  -d '{"client_name":"Muse","redirect_uris":["https://muse.ai/api/hatch/oauth/callback","https://agent.meta.ai/api/hatch/oauth/callback","https://platform.muse.ai/oauth/credential-test/callback"],"token_endpoint_auth_method":"none","grant_types":["authorization_code","refresh_token"],"response_types":["code"],"scope":"mcp:read mcp:write offline_access"}'

Copy client_id from the response and paste it into Muse. Leave the client secret empty.

The first two addresses are the consumer sign-in callbacks Muse uses in chat. The platform.muse.ai address is Meta's developer portal credential test. Register all three exactly, so the saved client matches the callback Muse sends.

If Muse registers the client itself and does not ask for a client ID, continue to sign-in. Use this same registration address for https://mcp.datadoe.com and for a white-label address on mcp.datadoe.com.

Step 3: Approve access

  1. Start the connection in Muse.
  2. Sign in with your DataDoe account for the Organization you want to use.
  3. Review the access request and approve it.
  4. Return to Muse.

Step 4: Check the connection

Start a new Muse chat and send:

Use DataDoe to list my connected Amazon accounts.

If Muse lists the accounts, the connection is ready. Then ask:

Use DataDoe to summarize my Amazon sales and advertising spend for the last 7 days.

Name the Amazon account or marketplace if you manage more than one.

Review tool requests before you approve them. Tools that perform Actions can change your Amazon data.

Connect with an MCP key

Use an MCP key when the connector can send a request header and you are not using OAuth.

  1. Open DataDoe MCP Integrations (opens in a new tab) and select New key.
  2. Create a personal key. Your role must allow Personal MCP keys.
  3. Copy the key before you close the dialog. DataDoe shows the full value only once.

Send the key in the datadoe-mcp-key header to the same server address, using Streamable HTTP. DataDoe reads an MCP key from that header. See Manage API keys, MCP keys, and service accounts.

If you need help

If Add custom connector is missing, ask Muse in chat to add a custom connector.

If sign-in fails, check the server address, allow pop-ups and redirects, and confirm the client ID is the value DataDoe returned. Leave the client secret empty.

If Muse connects but returns no data, check your DataDoe Organization and Amazon connections, then ask Muse to use DataDoe explicitly.

Helpful resources


DataDoe MCP resources

Check the following resources for more information: