scout

Scout lets an assistant evaluate an MCP server and verify scout attestations from inside the agent, read-only.

What you can do

• Evaluate an MCP server and check its scout attestations from within the agent session.
• Run read-only checks against hosts, with loopback allowed by default and other hosts gated by an allow list.
• No individual tools are listed in the manifest, so exact operations are not documented here.

When to use it

Useful when you want an assistant to assess another MCP server or confirm its attestations without leaving the conversation, and when you need a read-only check of a host before trusting it.

Requirements

Runs as a local stdio process, distributed as a container image run with Docker. The SCOUT_MCP_ALLOW environment variable lists hosts that scout_check may evaluate besides loopback; a leading dot allows subdomains, and inside the container loopback means the container itself. No authentication is declared.