The Agent Skills Directory

[SAFE]: The diagnostic framework relies on standard, non-malicious Docker commands (e.g., 'docker logs', 'docker inspect', 'docker ps') for investigating container states and application errors.
[SAFE]: Destructive resource management commands such as 'docker system prune' are accompanied by explicit warnings and guidance on safe usage (e.g., using filters for cleanup) to prevent accidental data loss in production environments.
[SAFE]: Network and permission troubleshooting steps use standard administrative tools (e.g., 'iptables', 'nslookup', 'chmod', 'usermod') in a manner consistent with legitimate system maintenance and troubleshooting.
[SAFE]: External references are limited to official Docker documentation and established, well-known diagnostic images (e.g., 'alpine', 'nicolaka/netshoot') for network analysis.
[SAFE]: The skill correctly identifies and warns against high-risk practices, such as exposing the Docker socket to containers or using privileged tools without caution.

docker-troubleshooting