The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the SN-Execute-Background-Script tool to run server-side GlideScript on a ServiceNow instance. Step 5 provides a script template where a variable issueNumber is intended to be replaced with user-supplied input. This construction creates a command injection surface where a malicious input could be used to break out of the string literal and execute unauthorized server-side code.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it retrieves and processes untrusted data from the description, short_description, and work_notes fields of GRC issue records. Malicious instructions embedded in these fields could influence the agent's behavior during the summarization process.
Ingestion points: ServiceNow tables sn_grc_issue, sn_grc_risk, and sn_compliance_control accessed in SKILL.md.
Boundary markers: None identified; the skill does not use delimiters or specific instructions to the AI to ignore embedded commands in the retrieved data.
Capability inventory: SN-Execute-Background-Script, Bash, SN-Query-Table as listed in the skill's tool definitions.
Sanitization: No evidence of data sanitization or input validation was found; raw data from the database is passed directly into the AI context and background scripts.
[DATA_EXFILTRATION]: The skill accesses highly sensitive organizational information, including risk ratings, failing compliance controls, and audit findings. While the skill's purpose is reporting, the mass extraction of this data into the AI's environment increases the risk of data exposure if the agent's session is compromised or the resulting summaries are misrouted.

issue-summarization