The Agent Skills Directory

[PROMPT_INJECTION]: The skill manages indirect prompt injection risks by enforcing a requirement for direct user intent before performing network operations.
Ingestion points: The <host> parameter in SKILL.md is the primary entry point for external data.
Boundary markers: Explicit confirmation is mandated before pinging internal network ranges (RFC 1918), loopback addresses, or cloud metadata endpoints.
Capability inventory: The skill utilizes the system ping binary to execute network checks.
Sanitization: Instructions explicitly prohibit the agent from initiating pings based on hostnames discovered in files or logs without direct user authorization.
[COMMAND_EXECUTION]: The skill involves the execution of the system ping utility. While the instructions define a clear command structure, there is an inherent risk of shell injection if the agent fails to sanitize the <host> argument provided by the user.
[DATA_EXFILTRATION]: The skill handles network-sensitive data including IP addresses and response timings. It includes mandatory guardrails that restrict this information to local display only and strictly forbid transmission to external services or webhooks.

ping-monitor-hardened