The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by processing external data from a ServiceNow instance while maintaining significant capabilities.
Ingestion points: Data enters the agent context through tools like snow_query_table, snow_discover_all_workspaces, and snow_uib_discover which read from ServiceNow system and application tables.
Boundary markers: The skill does not define specific delimiters or instructions to handle data retrieved from the ServiceNow instance as potentially untrusted.
Capability inventory: The skill has the ability to execute server-side scripts via snow_execute_script, perform table writes, and call internal ServiceNow APIs.
Sanitization: There is no explicit mechanism described for sanitizing retrieved data before it is processed or used to generate subsequent commands.
[DYNAMIC_EXECUTION]: The skill generates and executes dynamic content as part of its core development workflow.
It provides ES5 JavaScript templates (e.g., GlideRecord operations) intended for runtime execution on the target ServiceNow instance via snow_execute_script.
It manages the creation of UI Builder macroponents by assembling and writing JSON strings to the composition field of the sys_ux_macroponent table.

workspace-builder