The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses local Python scripts (scripts/lovstudio_llm.py and scripts/realtime_client.py) to interact with its API. All command-line arguments are structured, and the skill documentation provides clear examples of usage. No arbitrary command execution patterns were found.
[CREDENTIALS_SAFE]: The skill correctly manages sensitive information by instructing users to provide the LOVSTUDIO_API_KEY via environment variables. The scripts include redaction logic (_redact function in lovstudio_llm.py) to ensure that secrets are not accidentally leaked in error messages or logs.
[DATA_EXFILTRATION]: Network operations are directed to the author's domain (llm.lovstudio.ai), which is consistent with the skill's primary purpose. No evidence of sending sensitive user data to unauthorized third-party domains was found.
[PERSISTENCE]: The skill utilizes a local profile storage system (scripts/profile_store.py) to maintain session context (e.g., user preferences). This implementation is transparent, uses atomic file operations, and includes explicit checks to prevent sensitive keys or tokens from being stored in the durable profile.
[REMOTE_CODE_EXECUTION]: All primary logic is contained within the provided scripts, which rely almost exclusively on the Python standard library. The validation script requires PyYAML, but no code is downloaded or executed from remote servers at runtime.

lov-integrate-lovstudio-llm-skill