The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill describes tools, specifically snow_execute_script, that perform server-side execution of JavaScript on a ServiceNow instance. It details how scripts are evaluated using GlideEvaluator.evaluateString or executed via scheduled jobs (sysauto_script).
[INDIRECT_PROMPT_INJECTION]: The skill documents an environment with high capabilities (reading/writing to system tables and arbitrary script execution) that could be exploited if an agent interpolates untrusted user data into the scripts it generates.
Ingestion points: The script parameter of the snow_execute_script tool described in SKILL.md.
Boundary markers: The skill notes that scripts are often wrapped in IIFE patterns (function() { ... })() for execution.
Capability inventory: Full record management (snow_record_manage), database querying (snow_query_table), and server-side JavaScript execution (snow_execute_script).
Sanitization: Mentions a security_analysis feature that performs a regex scan for dangerous patterns (e.g., .insert(), .update(), eval()) and a validate_es5 flag.
[COMMAND_EXECUTION]: The documented tools facilitate the execution of arbitrary administrative commands within the ServiceNow platform environment, which could be misused if not properly constrained by the agent using the skill.

background-script-execution