The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by ingesting and processing untrusted data from ServiceNow. * Ingestion points: Data is fetched from sn_ot_incident, sn_ot_vulnerability, and sys_journal_field (work notes) tables in SKILL.md. * Boundary markers: While the skill uses a structured summary template, it lacks explicit boundary delimiters or 'ignore' instructions that would prevent the AI agent from acting on malicious commands hidden within incident fields like short_description or work_notes. * Capability inventory: The skill includes access to Bash and ServiceNow management tools (MCP/REST), creating a path for potential command execution if an injection is successful. * Sanitization: No validation, sanitization, or escaping of retrieved data is described in the procedure before the data is interpolated into the summary prompt.

ot-incident-summarization