The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from external network devices, which represents a potential attack surface for indirect prompt injection.
Ingestion points: Untrusted data enters the agent's context via the gnmi_get, gnmi_get_subscription_updates, and gnmi_compare_with_cli tools (SKILL.md).
Boundary markers: The instructions do not define clear delimiters or "ignore instructions" warnings to isolate external device data from the agent's core logic.
Capability inventory: The skill has significant capabilities, including modifying device state with gnmi_set and performing multi-vendor network operations.
Sanitization: There is no evidence of sanitization or structural validation to filter out potential injection strings from the gNMI or CLI outputs before they are processed by the agent.

gnmi-telemetry