The Agent Skills Directory

[COMMAND_EXECUTION]: The skill runs a local script scripts/mem-report.sh which executes several macOS system utilities including vm_stat, sysctl, ps, and awk. These commands are used strictly for diagnostic purposes related to memory monitoring.\n- [DATA_EXPOSURE]: The skill accesses system-wide information including the list of running processes and memory allocation metrics. While this exposes technical metadata about the host system to the agent, it is restricted to memory-related diagnostics and does not target sensitive user files or credentials.\n- [INDIRECT_PROMPT_INJECTION]: The skill accepts user-provided strings (like application names) to filter process groups in the report output, creating a potential ingestion surface for untrusted data.\n
Ingestion points: User queries about specific apps or RAM usage patterns in SKILL.md (e.g., scripts/mem-report.sh 'claude|Claude').\n
Boundary markers: No specific delimiters or instructions are used to isolate user-provided regex patterns from the command context.\n
Capability inventory: The skill possesses the capability to execute shell commands via the diagnostic script.\n
Sanitization: The input variable is double-quoted within the bash script (\"$1\"), which provides standard protection against shell metacharacter injection.

ram-check