The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses exec:bash to run system commands. Specifically, it executes pm2 commands with user-provided placeholders <file> and <name>. This creates a command injection vulnerability if these inputs are not sanitized to prevent shell metacharacters (e.g., ;, &&, |) before being passed to the shell.
[REMOTE_CODE_EXECUTION]: The skill invokes bun x gm-exec, which fetches and executes a package from a registry at runtime. This introduces a dependency on the availability and integrity of the external gm-exec package.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes process names and file paths that may originate from untrusted external tasks or requirements.
Ingestion points: Process names and filenames are passed from the gm-execute phase into the pm2 start command template.
Boundary markers: The instructions lack explicit boundary markers or instructions to treat these values as untrusted data.
Capability inventory: The skill has the capability to start, stop, delete, and view logs for system processes via pm2.
Sanitization: There is no evidence of sanitization or validation of the <file> or <name> parameters within the skill logic.

process-management