The Agent Skills Directory

[DATA_EXFILTRATION]: The skill defines mechanisms for sending data to external endpoints (webhooks, email providers). It mitigates exfiltration risks by emphasizing the use of the {{secret:name}} syntax for authorization headers and endpoints, ensuring credentials are not hardcoded.- [COMMAND_EXECUTION]: Includes a curl command within a documentation code block. This is provided as an example for users to manually test webhook connectivity and is not intended for automated execution by the agent.- [PROMPT_INJECTION]: The skill ingests untrusted event data (e.g., {{event_properties}}) for interpolation into output templates. While this creates a surface for indirect prompt injection, it is the expected functional behavior for a data activation service and is documented with standard variable substitution patterns.

rt-journey-activations