The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses docker exec to manage processes within containers, such as starting servers, checking health, and running benchmark scripts.
[COMMAND_EXECUTION]: Python scripts utilize subprocess.run() with argument lists to invoke the vllm bench serve CLI tool. This is a secure implementation practice that avoids shell injection vulnerabilities.
[SAFE]: The skill communicates with localhost:8000 to verify server status and retrieve model IDs. These operations are local and do not involve untrusted external domains or data exfiltration.
[SAFE]: A detection regarding piping curl output to python was determined to be a false positive; the implementation uses python3 -c with a hardcoded script to parse JSON data from the local benchmark service.
[SAFE]: No sensitive information disclosure, credential harvesting, or persistence attempts were identified in the codebase.

perf-test-flagos