The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from ServiceNow records and summarizes it into new tasks, creating a surface for indirect prompt injection.\n- Ingestion points: Data is ingested from the incident table (close_notes), kb_feedback table (comments), and search_log table (search_term) via the SN-Query-Table tool and ServiceNow REST API endpoints (SKILL.md).\n- Boundary markers: The instructions lack specific delimiters or boundary markers to isolate the retrieved text from the agent's internal instructions during processing.\n- Capability inventory: The skill possesses write capabilities through SN-Create-Record (creating kb_submission records), SN-Add-Work-Notes, and REST POST operations.\n- Sanitization: No explicit sanitization or filtering of the retrieved incident data or feedback comments is defined before the information is used to generate gap reports and new record descriptions.

gap-analysis