The Agent Skills Directory

[COMMAND_EXECUTION]: The skill generates Bash scripts that execute administrative system commands. Evidence: Phase 3 documentation includes examples using rm -rf on configuration directories (e.g., /etc/sysctl.d/*), sed for file editing, and systemctl for service control. These operations are localized to relevant system configuration paths and are necessary for the skill's purpose of verifying compliance rules.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface due to the ingestion of compliance rule content to generate shell scripts. 1. Ingestion points: rule.yml, OVAL XML files, and remediation scripts via MCP tools like mcp__content-agent__get_rendered_rule. 2. Boundary markers: No explicit delimiters or instructions to ignore embedded commands are used when reading and processing rule data. 3. Capability inventory: Generation of executable .sh files that are subsequently run by the Automatus framework. 4. Sanitization: No specific filtering or escaping is described; the agent is instructed to use values directly from analyzed rule content.
[SAFE]: The administrative operations and data processing described are core to the skill's legitimate function as a development tool for the ComplianceAsCode project. The use of destructive commands like rm -rf is limited to specific configuration paths for cleanup purposes within test environments.

create-test-scenarios