k7-total-security-deployment-automation

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.65). The skill’s runtime workflow can ingest outsider-authored free text via the configured threat-intelligence feed URL (e.g., threat_intel.sources: https://threat-intel.k7security.com/...), where fetched feed content is read as text and then provided to the agent/LLM for incident analysis.

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

• Potentially malicious external URL detected (high risk: 0.90). The installation steps explicitly curl installers from https://releases.k7security.com (e.g., https://releases.k7security.com/console/k7-console-16.0.1195-win-x64.msi, …-macos.pkg, …-linux-x64.tar.gz) and then run/execute them (msiexec/installer/install.sh), which means remote code is fetched and executed at runtime.

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

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The skill explicitly requires administrative/root privileges and contains commands that invoke sudo/msiexec, install system-wide software, and apply/modify security profiles on local and remote endpoints—actions that change the machine's state and require elevated access.