Is there a command-line tool to check an MCP server’s readiness?

mcp-readiness is a free, open-source command-line tool published by SaSame to npm. It grades a public Model Context Protocol server against the same spec-bound readiness criteria the Observatory uses — handshake correctness, tool description quality, safety annotations, anti-ghost-content checks, token efficiency and honest error behavior — plus a Claude/ChatGPT directory pre-flight check. It requires no API key, no account and no SaSame credential; the grading logic is public and inspectable in the source repository.

Run it directly with npx against any public MCP endpoint URL: npx -y mcp-readiness@latest gold-rush start <your-mcp-url> --json. This runs entirely on the caller’s machine or agent — it does not require SaSame to observe the endpoint first, and running it does not itself publish anything to the Observatory.

The CLI and the Observatory apply the same underlying readiness methodology, but they are not the same act: the CLI is a local, on-demand check anyone can run; an Observatory grade is a signed, dated observation SaSame itself performed and published. A CLI run is a useful way to predict or reproduce an Observatory grade before requesting one — it is not itself a substitute for a signed observation, and running the CLI does not update or create a public Observatory record.

The CLI reports a bounded technical readiness grade, not a safety, malware or endorsement verdict — the same boundary the Observatory itself states. It is independently open source: the exact grading code a reader runs is visible on GitHub, not a black box.