The Agent Skills Directory

[SAFE]: The skill follows security best practices by utilizing environment variables (e.g., %ENV{NETCLAW_USERNAME}, %ENV{NETCLAW_PASSWORD}) for device authentication within the pyATS testbed configuration, avoiding the risk of hardcoded credentials.
[SAFE]: All defined operations are restricted to read-only diagnostic commands (show version, show failover, show vpn-sessiondb, etc.), which prevents the agent from making unauthorized configuration changes to the firewall infrastructure.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external Cisco ASA firewalls. While this represents a potential injection surface, the risk is negligible as the skill lacks capabilities to perform file writes or external network requests based on the received output. This ingestion is a primary and necessary function for firewall monitoring.
[SAFE]: The skill instructions include explicit guardrails and workflows that prioritize human-in-the-loop verification (e.g., checking failover status before maintenance) and audit logging.

pyats-asa-firewall