The Agent Skills Directory

[COMMAND_EXECUTION]: The tests/test_litellm_health.py script uses subprocess.run to execute the local litellm-health tool during testing. This is a standard and safe use of subprocess execution within a controlled test suite.\n- [EXTERNAL_DOWNLOADS]: The skill documentation recommends installing the litellm package from PyPI. This dependency is fundamental to the skill and is sourced from a well-known project with provided security version requirements.\n- [DATA_EXFILTRATION]: The litellm-health script performs HTTP GET requests to a target proxy URL. It is a read-only probe designed for diagnostics and does not attempt to exfiltrate local system data or credentials to external domains.\n- [INDIRECT_PROMPT_INJECTION]: The skill involves the agent processing gateway output and logs. This operational workflow is standard for gateway management, and the documentation includes advice on redacting sensitive content to manage the risks inherent in processing external data sources.

litellm