The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Python scripts to invoke system diagnostic tools including leaks, xcrun, heap, and vmmap.
Evidence: scripts/capture_sim_memgraph.py and scripts/summarize_memgraph.py utilize subprocess.run to execute these commands.
Security Note: All commands are passed as argument lists rather than shell strings, which effectively prevents shell injection vulnerabilities. The tools invoked are well-known Apple developer utilities.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and parses text output from the leaks command-line tool.
Ingestion points: scripts/summarize_memgraph.py parses the output of leaks --list to identify memory addresses and type hints.
Capability inventory: The skill has the capability to write diagnostic artifacts to local directories and execute further diagnostic commands based on the parsed addresses.
Sanitization: The script uses specific regular expressions to extract structured data from the tool's text output, mitigating the risk of raw content influencing agent behavior directly. While a malicious application could theoretically craft memory metadata strings to appear in the tool output, this constitutes a low-risk diagnostic surface.

ios-memgraph-analysis