The Agent Skills Directory

[SAFE]: The skill uses curl to send requests to an official API endpoint (ai-factory.frevana.com). It allows for a configurable API base URL and token, following standard development practices.
[COMMAND_EXECUTION]: The skill uses a bash script to handle request logic and argument parsing. It includes an embedded Python script to safely construct a JSON payload and validate inputs (types such as bool, int, float, and strings), preventing shell injection vulnerabilities.
[CREDENTIALS_UNSAFE]: The skill handles a FREVANA_TOKEN for authentication. It provides a secure fallback by prompting for the token interactively if not provided in the environment or as a flag, and correctly uses it in the Authorization: Bearer header of the curl request.
[DATA_EXFILTRATION]: The skill interacts with a specific service (frevana.com). While it transmits a query to this service, the domain is the intended destination of the skill's primary function and is treated as a known service for this author context.

google-ai-mode