content-marketing

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill's companion-check instructs the agent on first activation to offer installing a remote package using the command "npx skills add AbsolutelySkilled/AbsolutelySkilled --skill " (also present as the concrete install command in the README), which would fetch and run remote code that can alter agent behavior.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.80). The companion check explicitly tells the agent to run shell commands that read local hidden directories and to offer installing missing skills via npx, which can expose local filesystem contents and cause the agent to modify the machine by installing software.