The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from ServiceNow tables (invoices, purchase orders, receipts) which could potentially contain content from external vendors or third parties.\n
Ingestion points: Data enters the agent context via SN-Query-Table and REST API calls to sn_proc_invoice, proc_po, proc_po_item, and proc_rec_slip_item.\n
Boundary markers: The instructions do not define specific delimiters or instructions for the agent to isolate potentially untrusted content within these fields.\n
Capability inventory: The skill has the ability to write to the ServiceNow database using SN-Create-Record and SN-Update-Record, and to execute shell commands through the Bash tool.\n
Sanitization: No explicit sanitization or validation of the retrieved text fields is described in the skill logic.\n- [COMMAND_EXECUTION]: The skill is configured to use the Bash native tool. In the provided procedure, this is used to perform legitimate REST API operations (e.g., GET, PATCH, POST) to the ServiceNow instance, which is a standard practice for this type of integration.

invoice-management