The Agent Skills Directory

[SAFE]: The skill is designed for auditing and reporting purposes. It follows security best practices by being read-only by default and requiring explicit user authorization for any write operations via the endpoint-update tool.
[COMMAND_EXECUTION]: The skill utilizes platform-specific tools including execute-sql, endpoint-materialization-status, endpoint-versions, and endpoint-update to perform its tasks. These tools are restricted to project management activities within the PostHog environment.
[DATA_EXFILTRATION]: The skill queries system metadata and logs, such as query_log and system.data_modeling_endpoints, to analyze endpoint performance and usage. There is no evidence of sensitive data being exfiltrated to external third-party services.

auditing-endpoints