The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses standard system utilities to perform network diagnostics, which is the intended and primary function.\n
Evidence: The instructions include commands like ping, curl, nslookup, dig, and openssl for Linux/macOS, and Test-Connection, Test-NetConnection, and Invoke-WebRequest for Windows (SKILL.md).\n
Context: These tools are used in a limited, read-only capacity to probe connectivity and verify TLS/DNS settings. The skill explicitly forbids dangerous flags such as --insecure or -k.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes network error messages which are derived from external server responses, presenting a theoretical attack surface for indirect instructions.\n
Ingestion points: The skill captures error text, target URLs, and failing commands from the environment and external services (SKILL.md).\n
Boundary markers: The skill defines comprehensive 'Safety Boundaries' that explicitly state safety rules must never be relaxed and prohibit the display of sensitive data like tokens or credentials.\n
Capability inventory: Capabilities include read-only shell commands for network probing (SKILL.md).\n
Sanitization: The skill mandates that internal hosts be replaced with placeholders and requires a classification and interpretation step before acting on any data, providing a layer of human and agent review.

network-troubleshooting