The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/main.cjs file uses execSync to run ripgrep commands. It constructs the command by joining the searchDirs array into a string. Since these directory paths are not sanitized for shell metacharacters (e.g., semicolons, backticks, or subshell syntax), a user or malicious actor providing crafted directory names could execute arbitrary commands. \n- [COMMAND_EXECUTION]: The script derives search terms directly from file names on the disk (nameWithoutExt). If a file is maliciously named to include shell-active characters, it could trigger unintended command execution when the script processes that file. \n- [COMMAND_EXECUTION]: The skill includes a destructive deletion capability (fs.unlinkSync). When combined with the command injection vulnerabilities in the search logic, this presents a risk where an attacker could manipulate the search results to delete arbitrary files on the system.

stale-module-pruner