configuring-experiment-analytics
Pass
Audited by Gen Agent Trust Hub on Apr 30, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a secure workflow for data interaction. It explicitly mandates calling
read-data-schemato verify the existence of event names before they are used in metrics, which serves as a validation step against processing untrusted or incorrect input names. - [DATA_EXPOSURE]: No hardcoded credentials, sensitive file paths, or unauthorized network operations were detected. The skill uses standard tool calls (
experiment-get,experiment-update,read-data-schema) consistent with its stated purpose of managing experimentation data. - [PROMPT_INJECTION]: The instructions are task-oriented and do not contain patterns typical of prompt injection, such as attempts to bypass safety filters or override system instructions.
- [REMOTE_CODE_EXECUTION]: There are no patterns involving remote script execution, package installation, or dynamic code evaluation. The skill relies entirely on predefined platform tool calls.
Audit Metadata