skills: why it is not safe to use
We found broad external tool access and a remote installer pattern that increases security and privacy review needs before use.
What we found
Our review did not find hidden instructions, hardcoded secrets, obfuscated content, or clearly destructive commands in the sampled files. The main concern is how this skill is intended to be used. The repository promotes installing via a remote script piped directly into `bash`, which reduces an admin's chance to inspect what will run before execution. We also found that the `mcp-hub` skill is designed to connect Claude to external MCP servers and a wide range of connected systems, including filesystems, databases, APIs, and named third-party tools.
The reviewed materials also describe integrations such as Google Drive, GitHub, Slack, browser automation, databases, and local filesystem access. Once those tools are connected, the skill may be able to access or act on sensitive user or company material through those services, depending on the permissions granted.
What to do instead
Prefer skills that can be reviewed locally before installation and avoid piping remote installers into a shell. If you evaluate this skill, limit it to a test environment, connect only the minimum required tools, and use narrowly scoped credentials and permissions. Review each MCP server or third-party integration separately before enabling access to internal files, messages, repositories, or databases.
Source: https://github.com/claude-office-skills/skills
We report what our security review found at the time we checked, with the goal of keeping people safe. Projects change; if a maintainer has since fixed this, we are glad to recheck it. Email hello@agentpod.com.