Outside eyes for the sites your agent ships.

Your agent can read every file in the repository. It cannot see the certificate a browser receives, the header a CDN dropped or the DNS record that did not propagate. Guardr checks all of that from outside, grades it from A to F and tells the agent what to fix.

1 - What an agent gets

Three tools, over the Model Context Protocol, at https://api.guardr.io/mcp. Each one runs as the holder of your API key, so an agent reads exactly what the REST API returns for the same account.

Each tool returns the same fields as the REST API route it matches. The reference names the routes.

Every tool result carries the allowance it was served under in its _meta.rate_limit field: limit, remaining, reset and retry_after. Listing the tools is the exception: that call spends the allowance without reporting it, so its result carries no _meta.rate_limit.

2 - Connect in one line

Claude Code

One command registers the endpoint. Run it in any project.

claude mcp add --transport http guardr https://api.guardr.io/mcp --header "X-API-Key: your-key-here"

Cursor

Add one server to the MCP configuration file Cursor reads.

{"mcpServers":{"guardr":{"url":"https://api.guardr.io/mcp","headers":{"X-API-Key":"your-key-here"}}}}

Where Guardr is listed

A client that installs from a registry rather than from a command finds the same endpoint in two places.

3 - What Guardr does not do

Guardr proposes, it does not apply. No tool changes a site, a repository or a deployment. Guardr reads what any visitor can read. An agent reads a grade and a list of fixes, then decides what to do with them the way it decides anything else.

4 - Plans

The endpoint is in both plans, on the same key and the same allowance as the REST API. One site is free with no card. Every site after the first is $3.00 each per month. The plans are on the pricing page and the endpoint is documented in full on the MCP endpoint page.

Guardr keeps watch.