The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/find_ai_file.py uses subprocess.run to execute the ripgrep (rg) utility for fast text searching. The command is constructed safely using a list of arguments without shell=True, and re.escape() is applied to all user-controlled search terms to prevent injection.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted local transcript data (.jsonl files) to identify file paths. While this presents an attack surface for indirect prompt injection, the risk is minimized by the skill's focus on local file system observation (stat) and robust sanitization of query terms.
[DATA_EXPOSURE]: The skill accesses local AI session directories and user-specified roots to locate files. All metadata and indices are stored locally in the user's cache directory with restrictive 0600 permissions. No network exfiltration or unauthorized access was detected.

lov-search-file