The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process incident data from ServiceNow, which represents a vector for indirect prompt injection if the ingested content contains malicious instructions.\n- Ingestion points: Data is retrieved from the incident table (specifically short_description and description fields) using the snow_record_manage and snow_query_table tools.\n- Boundary markers: The skill does not define specific delimiters or guidelines to help the agent distinguish between its own instructions and the data processed from incident fields.\n- Capability inventory: The skill utilizes the snow_execute_script tool for running JavaScript code on the ServiceNow server and snow_record_manage for record updates, providing a powerful set of capabilities that could be targeted by injected instructions.\n- Sanitization: There is no evidence of logic to sanitize or validate external data before it is incorporated into tool arguments or scripts.

incident-management