Asking in your editor: DomainWarn as an MCP server

Asking in your editor (MCP)

DomainWarn is an MCP server itself: one address, one token, nothing to install. Editors and assistants that speak the Model Context Protocol – Claude Code, Claude Desktop, Cursor, Zed, VS Code – then answer questions like “Which client domains are critical right now?”, “Why is the DMARC check for example.com red?” or “Check example.com” with your real data, without switching tabs.

The server offers nine tools in three tiers. Without a token the free checks run: check_domain (the full domain check with a score) and run_check (a single check such as SPF or SSL) – no account needed. With a token from Settings → API you also get list_domains, get_domain, list_incidents and get_dns_history.

Writing tools exist only for a token with write scope: the assistant may then use add_domain, check_now and acknowledge_incident. With a read token they never even appear – so nobody adds domains by accident and blows past your plan limit.

Setting it up

# Claude Code
claude mcp add --transport http domainwarn https://api.domainwarn.com/api/v1/mcp \
  --header "Authorization: Bearer dw_…"

# without a token: the free checks only
claude mcp add --transport http domainwarn https://api.domainwarn.com/api/v1/mcp

In applications that read a file with mcpServers, add an entry with type: "http", url: "https://api.domainwarn.com/api/v1/mcp" and the token under headers. Revoking access works like anywhere else: delete the token under Settings → API, and the server falls back to the free checks.

MCP in your editor

Asking in your editor (MCP)