Email Spam Tester

What Email Spam Tester MCP server does

The Email Spam Tester MCP server gives an agent a way to test how an email is received and evaluated. It connects to the hosted service at https://email-spam-tester.com/mcp, which uses streamable HTTP and does not require an API key or account.

A test begins with a disposable recipient address. The agent must send a real message to that address from the email platform under evaluation. This matters because many checks depend on headers added by the sending platform; a message sent from a local mail client may not represent a production campaign.

The resulting report covers 42 checks, including SPF, DKIM, DMARC, 21 blocklists, two spam engines, and Gmail and Yahoo bulk-sender requirements. Failed and warning checks include evidence, links to relevant RFCs, and a fix plan. The report also provides scores, but an incomplete analysis should be treated as optimistic.

How it works

Use get_test_address(lang="en") to reserve an address. The response includes a slug and a report URL. The language setting controls the fix plan shown to a person viewing the report; the per-check lines returned by the tools remain in English.

After sending the message, call wait_for_report(slug, timeout_seconds=150). This operation waits for analysis to finish, so an agent does not need to implement polling or a sleep loop. Use get_report(slug) when a report already exists and should be read without waiting.

get_message_source(slug) returns the delivered message, including headers in wire order, both body versions, and the raw source. A skip status means a check did not apply, while error means the check could not run and is excluded from the score. The complete field indicates whether every possible check finished.

The slug grants access to the report data, including the subject, sender, bounce address, and complete source. It should be handled as a capability and used only for mail that the tester is authorized to inspect.

Setup and configuration

No server-side installation is needed because the MCP endpoint is hosted by the service. Configure an MCP client with the URL https://email-spam-tester.com/mcp.

Claude Desktop can use a direct streamable HTTP entry in claude_desktop_config.json:

{
  "mcpServers": {
    "email-spam-tester": {
      "url": "https://email-spam-tester.com/mcp"
    }
  }
}

Claude Code can be configured with:

claude mcp add --transport http email-spam-tester https://email-spam-tester.com/mcp

Clients without direct streamable HTTP support can use mcp-remote through an MCP configuration with npx -y mcp-remote https://email-spam-tester.com/mcp. The repository includes a complete Claude Desktop example and an MCP registry manifest. The MIT license applies to the repository, not to the hosted service.

Tools and capabilities

The Email Spam Tester MCP server exposes four tools:

• get_test_address: Reserves a disposable address and returns its slug and report URL.
• wait_for_report: Waits for analysis and returns scores, failing and warning checks, evidence, and the fix plan.
• get_report: Retrieves an existing report without waiting.
• get_message_source: Retrieves the delivered message and its raw source.

Together, these tools support testing from message submission through report review and source inspection. The agent remains responsible for sending the message between address creation and report retrieval.

Limitations and notes

The service evaluates the message that actually reaches the disposable address, not an abstract campaign configuration. Results can therefore differ when the sending platform, routing, or message construction changes.

A report with complete: false is missing one or more checks and should not be treated as definitive. Skipped checks are not passes, and errored checks do not contribute to the score. The slug also exposes sensitive message data to anyone who possesses it, so it should not be shared casually or used to inspect another person’s email without permission.