The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses sudo and platform-specific installers (such as msiexec and dpkg) for the installation of security agents and management consoles. These high-privilege operations are appropriate for the intended purpose of deploying system-level endpoint protection.
[EXTERNAL_DOWNLOADS]: Fetches deployment archives and installation packages from the official vendor domain download.k7computing.com. These downloads are a standard part of the software distribution workflow.
[CREDENTIALS_UNSAFE]: Accesses sensitive certificate files at paths like /etc/k7/admin.pem for console and API authentication. This is consistent with local certificate-based authentication requirements.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the analysis of threat events by passing data to OpenAI and Claude models. This creates a potential surface where malicious strings within threat descriptions could attempt to influence the AI's analysis. Ingestion points: Threat event JSON payloads received from the management API or webhooks. Boundary markers: None identified in the sample configuration. Capability inventory: Ability to apply security profiles via k7-console and send network notifications to Slack. Sanitization: No explicit sanitization of event data is demonstrated in the provided code snippets.

k7-total-security-deployment-framework