​
Overview

Safeturf publishes a remote MCP server that Claude can connect to as a custom connector. Once connected, your team can ask Claude questions like:
  • “What monitoring items need attention?”
  • “Does <business name> have a consumer-sentiment flag?”
  • “Which data sources produced that verdict?”
Claude reads the same monitoring data you see in the portal — scoped to your partner, and read-only. It cannot change verdicts, resolve items, or edit anything in Safeturf.
Everyone signs in with their own Safeturf login, and only ever sees their own partner’s businesses. Adding the connector for your organization shares the connector, never anyone’s credentials or access.

​
What Claude can and cannot see

Claude can readClaude cannot access
Monitoring items needing attention (flags, expirations)Social security numbers, dates of birth, bank/account numbers
Each check’s verdict, reasoning, and the data sources behind itRaw responses from third-party data providers
The underwriting decision on file for each checkDocument contents (only filenames, types, and dates)
Document metadata for a businessAny other partner’s businesses

​
Adding the connector for your organization

On Team and Enterprise plans, only an organization Owner can add a custom connector. If you are a Member, the option will not appear — ask your Owner to complete these steps.
1

Open organization settings

Go to Organization settings → Connectors.
This is not the same as the personal Customize → Connectors page. Adding it there only connects your own account, not your team’s.
2

Add a custom web connector

Click Add, choose Custom, then Web.
3

Enter the Safeturf server URL

https://monitoring.safeturf.ai/api/mcp
4

Set the OAuth Client ID

Open Advanced settings and set:
claude-mcp
Leave the client secret blank — this is a public client that uses PKCE.
Do not skip this step. Safeturf’s sign-in service does not allow automatic client registration, so without the client ID you will get an error like “Couldn’t register with Safeturf’s sign-in service.” Entering the client ID manually is the intended setup, not a workaround.
5

Click Add

The connector is now available to everyone in your organization — but nobody is connected yet. Each person authorizes individually in the next section.

​
Connecting as a team member

Once an Owner has added the connector, each person enables it for themselves:
1

Open your connector settings

Go to Customize → Connectors and find the Safeturf connector, labeled Custom.
2

Click Connect

You will be redirected to the Safeturf sign-in page. Sign in with your normal Safeturf account and approve access.
3

Ask Claude something

Start a new chat and ask “What Safeturf monitoring items need attention?” If the connector is working, Claude returns your flagged businesses.

​
Individual accounts

If you are on a Pro or Max plan rather than a Team or Enterprise organization, you can add the connector yourself: go to Settings → Connectors → Add custom connector and follow the same steps — the URL and OAuth Client ID are identical.

​
Troubleshooting

The OAuth Client ID is missing. Edit the connector, open Advanced settings, set the client ID to claude-mcp, leave the secret blank, and connect again.
Your Safeturf account may not have a partner assigned. Claude will say something like “your SafeTurf login has no partner assigned.” Contact your Safeturf administrator — this is a permissions issue on the account, not the connector.
On Team and Enterprise plans only Owners can add custom connectors. Ask your organization Owner to add it, then connect from Customize → Connectors.

​
Questions to try

Triage the queue

“What monitoring items need attention?” — returns your flagged businesses, most severe first.

Investigate one merchant

“Does <business name> have any open flags?” — every check’s current verdict.

Understand a verdict

“Why is <business name> flagged, and which sources said so?”

Separate new from accepted risk

“Has anything changed since we approved <business name>?”
Claude also knows what was accepted at underwriting, so it distinguishes genuinely new risk from conditions your team already reviewed and approved.