The Agent Skills Directory

[SAFE]: The instructions promote secure and reliable behavior by mandating that the agent call read-data-schema to discover legitimate project events before suggesting any metrics. This 'verify-before-suggest' pattern mitigates the risk of hallucination and ensures the agent operates on factual project data.
[SAFE]: The skill uses established platform tools (experiment-get, experiment-update, read-data-schema) for their intended purposes without any patterns indicating data exfiltration, credential harvesting, or unauthorized privilege escalation.
[SAFE]: No obfuscation, prompt injection, or remote code execution vectors were found. The skill content is transparent and consistent with its stated purpose of managing experiment configuration.

configuring-experiment-analytics