The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides instructions for the agent to ingest and process data from external ServiceNow tables, which is an inherent surface for indirect prompt injection.
Ingestion points: The skill utilizes SN-Query-Table to fetch records from several tables including sysapproval_approver, sc_ic_aprvl_type_defn, and sys_user_delegate (SKILL.md, Steps 2, 4, 8, and 11).
Boundary markers: There are no explicit delimiters or instructions provided to the agent to disregard potential prompts embedded within the ServiceNow record fields (such as comments or descriptions).
Capability inventory: The skill possesses the ability to perform write operations via SN-Create-Record and SN-Update-Record (SKILL.md, Steps 3, 5, 6, 9, and 10), which could be triggered by processed input.
Sanitization: The instructions do not specify any sanitization, filtering, or validation logic for the data retrieved from the ServiceNow instance.

approval-workflows