The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill relies on SN-Execute-Background-Script to perform data aggregation and impact analysis.
Evidence: Steps 4, 5, and 8 define templates for server-side JavaScript that are executed via the MCP tool.
Risk: This tool allows arbitrary code execution on the target ServiceNow instance. If the agent populates script variables with unsanitized data from external regulatory bodies, it could lead to server-side script injection.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from regulatory alerts to drive compliance actions.
Ingestion points: Data is pulled from the description and short_description fields of the sn_regulatory_alert table (SKILL.md, Step 5).
Boundary markers: Absent. The skill does not instruct the agent to treat this data as untrusted or provide delimiters.
Capability inventory: The agent can create/update records, run background scripts, and execute shell commands.
Sanitization: Absent. There is no logic to filter or escape content retrieved from external sources before it influences the agent's recommended actions or task descriptions.
[COMMAND_EXECUTION]: The skill utilizes Bash to interact with REST APIs for data manipulation.
Evidence: Native Bash is listed in the manifest and used in Steps 1, 2, 6, and 7 for ServiceNow Table API requests.

regulatory-alert-analysis