The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill includes the SN-Execute-Background-Script tool, which allows for the execution of arbitrary server-side JavaScript (GlideScript) on the target ServiceNow instance. This is demonstrated in 'Example 3: Bulk Reply Analysis', where a script is passed to the tool for execution. This capability can be used to bypass platform security controls, access sensitive data, or modify system configurations.
[DYNAMIC_EXECUTION]: The use of SN-Execute-Background-Script involves the generation and execution of code at runtime. The skill encourages the agent to use this tool for 'complex matching logic or bulk recommendation analysis', which involves dynamic script construction.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from multiple sources to generate chat recommendations.
Ingestion points: Data is retrieved from the interaction table (chat history) in Step 2 and the incident table (short_description, description) in Step 1.
Boundary markers: The skill lacks explicit instructions to treat retrieved data as untrusted or to ignore embedded instructions within the chat history or incident fields.
Capability inventory: The agent has the ability to write back to the system via SN-Update-Record, SN-Add-Work-Notes, and execute scripts via SN-Execute-Background-Script.
Sanitization: There is no evidence of sanitization, escaping, or filtering of the external content before it is processed or interpolated into reply templates.
[PRIVILEGE_ESCALATION]: The tool SN-Execute-Background-Script is traditionally restricted to administrative roles (e.g., security_admin or admin) in ServiceNow due to its power. The skill, however, lists lower-level roles such as itil and agent_workspace_user in the prerequisites, suggesting a potential mismatch where the tool might grant the agent higher effective privileges than the intended user role.

itsm-chat-reply-recommendation