About Verune

Verune is a catalog of public AI agent tools and models: MCP servers, agent skills, Claude Code plugins and model repositories. It finds them in public registries, downloads their published code where it can, and examines it with static analysis. Nothing is ever executed. The result is a report per tool that says what its code can do and what changed between versions.

Who runs it

Verune is built and run by Netanel Zelig, a solo founder with a background in cyber security and engineering. It is an independent project, not a registered company.

Articles on the blog are signed “Verune research”. They are written from Verune's own data, and each one says when the data was taken.

How far to trust it

Every finding comes from automated rules, and no person has reviewed each one. Each finding says how sure we are that the code really does what it describes. Where we have measured a type of finding, we publish how often it was right, and we withdraw the number when the rule changes. When our own checks turned out to be wrong, we said so and fixed them. The methodology explains all of this.

A finding is a place to look, not a verdict. A tool can run shell commands or call the network for good reasons, and a missing finding is never a safety certification.

Contact

Email nati.zelig@gmail.com.

• You maintain a tool and think a finding is wrong. Send the tool's page and what the code really does. We will look and correct it if we were wrong.
• You found a problem in Verune itself. Please report it privately first.
• Press, research or anything else. Same address.