The Agent Skills Directory

[COMMAND_EXECUTION]: The hub coordinates the execution of various local administrative tools and scripts, including python3 scripts/linux_admin_activity.py for logging and reporting, and install-skills-bin for setting up core utilities.
[PRIVILEGE_ESCALATION]: The skill instructions involve the use of sudo for high-privilege tasks such as binary installation and service configuration validation. It also requires the agent to operate within protected system paths like /etc/nginx and /usr/local/bin.
[DATA_EXFILTRATION]: The framework includes patterns for syncing data to external storage, specifically mentioning rclone integration for Google Drive backups.
[INDIRECT_PROMPT_INJECTION]: The routing mechanism relies on user-provided descriptions of symptoms and tasks to select specialized workflows.
Ingestion points: User request fields for server roles and observed symptoms identified in SKILL.md.
Boundary markers: Absent; no specific markers are used to encapsulate or delimit user-provided content.
Capability inventory: Broad system access including script execution, privileged command execution via sudo, and repository management.
Sanitization: Absent; the instructions do not specify validation or escaping of user input during the routing phase.

linux-sysadmin