What 67 MCP servers do at startup, and what it means for your launch

What happened

A study published at mcp.brydge-ai.com looked at 67 npm packages. Each was installed, started in a room with no route to the internet and no real credentials, asked for its list of tools, and compared against its own README. No tool was called.

The headline figures, according to the study's own page: 24 servers offered a tool their README does not name, with 1,143 tools offered in total. The page notes that finding nothing is not the same as a clean sweep.

The page is careful about what this does not show. It is not a ranking, it is not calibration-backed, and it does not test whether any tool does what it says. The rule reads only the README that shipped in the package, so a server that documents its tools in a reference file beside it shows every tool as unnamed, and 10 of the 24 servers have a README under three kilobytes. In the page's words, a large count is a fact about the rule as much as about the package.

What it means for a founder about to launch

You do not need to be building an MCP server to take something from this. The study reads one document, the README, and compares it with what the product does at startup. Anything the product does that the document does not say shows up as a gap, whether or not the gap is harmful.

There are three practical habits here. First, list every capability the product exposes, not just the headline one. Second, name every outside service it contacts, so nobody has to find out by watching network traffic. Third, if your detail lives in a separate reference file, say so in the first screen of the README and link it, because a quick reader will not go looking.

Short is fine; silent is not.

None of this decides whether people want your product. It is about making the first minute of looking at it go smoothly.

Where LaunchBeam fits

When you launch, your listing is the short, public statement of what the product is, and it should say what the README says. Write both in one sitting and check they agree.

On LaunchBeam, a free launch is reviewed and listed within 14 days, with a verified dofollow badge link back to your site. That link points readers at your real home, so make sure the README, the site and the listing tell the same story. We cannot promise traffic or upvotes. What you control is whether the people who do arrive find a product that matches its description.

Before you submit, read your own README the way the study did: list what it names, list what the product actually does, and fix the difference.