Show HN: ModVet – vet your Claude Code mods before you install them (github.com)

ModVet has launched a crucial new tool for the AI/ML community, allowing developers to vet their Claude Code modifications before installation. This tool scans for potential vulnerabilities, including prompt injection patterns and malicious capabilities, all while operating offline and in just milliseconds. By offering a deterministic approach to identifying security risks—such as shell execution and file access—ModVet enhances the safety of mod installations without relying on cloud services or external accounts. Significantly, ModVet supports various output formats, including JSON and SARIF, making it compatible with GitHub's code scanning tools. Its update drift feature allows re-audits to target only changes after an update, effectively reducing the attack surface by highlighting new vulnerabilities. With extensive mapping to the OWASP MCP Top 10 security risks and a notable reduction in false positives since its initial version, ModVet serves as an essential defense mechanism in mod development, ensuring that security is prioritized in the AI ecosystem. Developers can easily install ModVet via pip, requiring only Python 3.9 or newer, streamlining integration into existing workflows.