> ## Documentation Index
> Fetch the complete documentation index at: https://docs.birdeye.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Troubleshooting

> Common issues and how to resolve them when using the Birdeye MCP Server.

## Connection Issues

### "Failed to connect to MCP server"

The AI client cannot reach `https://mcp.birdeye.com/mcp`.

* Verify the URL is entered exactly as: `https://mcp.birdeye.com/mcp`
* Check that your network allows outbound HTTPS connections
* Try disconnecting and re-adding the connector in your AI client

### "MCP server not responding"

The server may be temporarily unavailable. Wait a few minutes and retry.

***

## Authentication Issues

### "Authorization failed" or OAuth redirect not completing

* Ensure you are signing in with the correct Birdeye account credentials
* Check that your browser allows pop-ups from `mcp.birdeye.com`
* Try the connection flow in an incognito/private browser window

### "Token expired" or tools stop working after some time

OAuth access tokens have a limited lifetime. Most AI clients automatically refresh tokens in the background. If your client does not:

1. Disconnect the Birdeye connector from your AI client settings
2. Reconnect and complete the OAuth flow again to issue a new token

### Using an MCP client other than Claude or ChatGPT

Non-standard clients require a pre-registered redirect URI. Contact Birdeye Support with your client's redirect URI to get it authorized.

***

## Tool Errors

### "Tool returned no data" or empty response

* The date range or filters may be too restrictive — try widening the date range or removing filters
* Your account may not have data for the requested category (e.g., no surveys created yet)
* Some tools require `businessNumbers` from `get_child_locations` — confirm that step completed first

### "Missing required parameter"

Check the tool's parameter table in this documentation. Required parameters must be provided before the tool can execute.

### Tools work but return stale data

The Birdeye MCP Server returns live data on every call — there is no caching layer. If data appears stale, the underlying Birdeye account data may not have updated yet. Review ingestion times in your Birdeye dashboard.
