Overview
The Valimail MCP connection lets you ask Claude questions about your Valimail data in plain language. Claude reads your DMARC, sender, and domain reporting through the Valimail MCP (Model Context Protocol) server and answers in the chat.
Example uses:
Summarize DMARC pass/fail trends for a domain over the last week
List senders that are failing authentication and suggest next steps
Pull a quick domain health snapshot before a customer or leadership review
The connection is read-only. Claude can view reporting data your Valimail user can already see; it cannot change configuration, senders, or DNS.
Labs feature: The Claude integration is currently offered as a Labs feature on the Enterprise package. Claude is the first supported AI client; others (such as ChatGPT and Microsoft Copilot) are planned.
Requirements to Connect
Enterprise Package
Be an account owner
Have 2-factor authentication or Single Sign-on enabled
Connect Valimail to Claude
Setup takes two steps: a Claude admin adds the Valimail connector once, then each user connects their own Valimail login.
Step 1: Enable the connection (Claude admin)
In Valimail, go to Account Settings → Integrations and open the Claude card.
Copy the Valimail MCP server URL and OAuth Client ID shown in the setup instructions.
In Claude, go to Settings → Connectors, choose Add custom connector, name it "Valimail", and paste the URL and OAuth Client ID.
Save. The connector is now available to members of your Claude organization.
Step 2: Connect your Valimail account (each user)
In Claude, open Settings → Connectors and click Connect next to Valimail.
You're redirected to Valimail. Sign in (SSO works as usual) and approve Claude's read-only access.
You're returned to Claude. Enable the Valimail connector in a chat and start asking questions.
Once connected, the Claude card in Valimail shows Manage instead of Connect, and your connection appears in the connections table.
What you can ask
The MCP server exposes Valimail's reporting data as tools Claude can call.
Data area | Example prompt |
DMARC results | "How did DMARC pass rates change for example.com over the last 7 days?" |
Senders | "Which senders are failing SPF or DKIM, and what should I fix first?" |
Domains | "Give me a health summary of all my domains, sorted by risk." |
Tips:
Name the domain and time range for the most precise answers.
Claude only returns data available in your Valimail reports; it won't invent senders or domains it can't see.
Security and data handling
Your connection, your consent. You choose to connect Valimail to your organization's Claude. Data is shared only with the Claude organization you authorize, under your agreement with Anthropic.
By logging into Valimail and giving permission, Claude receives access to your account's reports and events via the OAuth 2 protocol. That permission is associated with your user, and can be revoked at any time. There are no cross-account credentials.
Read-only. The server wraps Valimail's Reports API, which only supports read (GET) requests.
Audited. Requests made through Claude are logged like any other Reports API call (user role, IP address, status).
Access tokens issued to Claude are valid for 24 hours and refresh automatically while the connection is active.
Manage or revoke access
Go to Account Settings → Integrations → Claude → Manage to see each connection's name, created date, and last used time. To revoke a connection, click Delete next to it; Claude loses access immediately. You can also disconnect from Claude's Connectors settings.
Note: Last used shows when Claude last received an access token, so it can lag actual usage by up to 24 hours.
Will Claude train on my Valimail data?
That depends on your Claude plan's data terms. Claude Team and Enterprise plans don't train on customer content by default.
Can Claude make changes in Valimail?
No. The tools provided to Claude in this connection are read-only. In the future, we may add account-editing tools to the connector, but your approval will be required before Claude is able to use them.