The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted text from incident descriptions which could potentially contain malicious instructions intended to influence routing or data updates.
Ingestion points: The short_description and description fields are fetched in Step 1 and processed for keyword signals in Step 2.
Boundary markers: The skill does not define specific delimiters or instructions for the agent to ignore embedded commands within the incident text.
Capability inventory: The skill possesses write capabilities including updating incident records (SN-Update-Record) and adding work notes (SN-Add-Work-Notes).
Sanitization: The procedure does not explicitly mention sanitizing or escaping incident content before performing content analysis.
[EXTERNAL_DOWNLOADS]: The skill interacts with external ServiceNow REST API endpoints to manage incident data.
The skill fetches historical patterns and valid categories from standard ServiceNow table APIs (/api/now/table/sys_choice, /api/now/table/incident).
These operations target the organization's official ITSM infrastructure for legitimate automation purposes.
[COMMAND_EXECUTION]: The skill lists Bash as a native tool and provides curl-style REST API examples for manual implementation.
These commands are used to interact with the vendor's expected API endpoints and do not exhibit suspicious patterns like obfuscation or unauthorized access.

predict-assignment