Prerequisites
- Access to an active Qwairy team on a Starter plan or higher, including paid Agency plans
- A client that supports remote MCP servers over Streamable HTTP
- To query monitoring data: at least one accessible active brand with generated responses
Choose a client
- Claude Desktop / Claude.ai
- ChatGPT
- Cursor
- Claude Code
- VS Code / Copilot
- Windsurf / Devin Local
- Other Clients
- Go to Settings → Connectors
- Click Add custom connector
- Enter a name (e.g.
Qwairy) and the server URL:
- Click Add, then authorize the connection when prompted
Custom connectors are available on supported Claude plans. Team and Enterprise workspace owners may need to enable or add the connector for the organization. See Anthropic’s custom connector guide.
Authorize the connection
After adding the server, your client will initiate the OAuth flow:- A browser window opens with the Qwairy sign-in page
- Sign in with your Qwairy account
- Authorize the connection
- You’re redirected back: the connection is active
Your credentials are never shared with the AI client. Authentication uses OAuth 2.1 with PKCE: only a secure, scoped access token is exchanged. Learn more about OAuth.
Query your data
Try these example prompts:Troubleshooting
”Server not found” error
- Use the root endpoint
https://mcp.qwairy.co; do not append/mcp - Check that your config file is valid JSON
- Restart your AI client after editing the config
”Authentication failed” error
- Check that you have a Starter plan or higher
- Ensure you’re signing in with the correct Qwairy account
- Try disconnecting and reconnecting
”No data returned”
Make sure you have:- At least one brand with status “Live”
- Monitoring enabled with generated responses
- Recent data (within the last 30 days by default)
Client doesn’t support remote MCP?
Prefer a client with native Streamable HTTP and OAuth support. If your organization approves a bridge, pin its version and follow that client’s documented OAuth callback procedure.Rate limits
The MCP server currently allows 300 requests per minute and 3,000 requests per day per access token. HTTP429 responses expose X-RateLimit-Limit, X-RateLimit-Remaining, and X-RateLimit-Reset for the per-minute window; successful responses do not expose quota headers. The JSON error message’s retry delay is derived from that same per-minute reset. Apply exponential backoff, and if requests remain limited after the reset, stop automated retries and resume later because the daily cap may be exhausted. These service limits can evolve; treat the documented values as current policy and 429 headers as burst-window guidance, not a complete daily-quota signal.
Measurement tools and scopes
Tools that read off-AI analytics (Search Console, Bing, referrer traffic, AI revenue, crawler activity) live under the Measure category and require theread:measure scope. If you connected before this scope existed, disconnect and reconnect to pick it up: scopes are never expanded silently on an already-issued token. See Measure Tools.
Next steps
Tools Reference
Learn what each tool does and its parameters
Example Conversations
See real-world usage examples
Combine with Other Tools
Connect Qwairy with your SEO, CMS, and project management tools
OAuth Setup
How authentication works under the hood

