integration-hub
Pass
Audited by Gen Agent Trust Hub on Feb 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONCREDENTIALS_UNSAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The tool 'snow_execute_script_with_output' enables the execution of arbitrary JavaScript (ES5) logic on the target ServiceNow instance.
- [CREDENTIALS_UNSAFE]: The skill provides code templates for the programmatic creation and management of sensitive ServiceNow records, including 'basic_auth_credentials' and 'sys_alias' (Connection/Credential Aliases).
- [DATA_EXFILTRATION]: The skill facilitates the creation and configuration of REST steps in ServiceNow actions, which are used to transmit data to external HTTP endpoints.
- [PROMPT_INJECTION]: Potential surface for Indirect Prompt Injection:
- Ingestion points: Action inputs such as 'summary', 'description', and 'priority' defined via the 'addActionInput' helper.
- Boundary markers: No explicit delimiters or boundary markers are defined in the provided script templates to separate untrusted data from instructions.
- Capability inventory: Access to the 'snow_execute_script_with_output' tool and the ability to define external REST calls through 'sn_ih_step_config'.
- Sanitization: The provided REST step templates demonstrate direct interpolation of inputs (e.g., '${inputs.summary}') into JSON payloads without evidence of escaping or validation.
Audit Metadata