The Agent Skills Directory

[COMMAND_EXECUTION]: Uses sudo to perform administrative operations, including modifying iptables firewall rules, viewing kernel logs with dmesg, and accessing restricted log files in /tmp.
[COMMAND_EXECUTION]: Employs docker exec and docker logs to interact with containers, allowing the execution of arbitrary commands within the awf-squid and awf-agent environments.
[COMMAND_EXECUTION]: Instructs the agent to run a local cleanup script (./scripts/ci/cleanup.sh) and remove docker networks and containers.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface by reading external traffic logs.
Ingestion points: Network access logs are read from /var/log/squid/access.log using cat, grep, and awk (SKILL.md).
Boundary markers: No delimiters or warnings are used to prevent the agent from interpreting log content as instructions.
Capability inventory: The skill can execute shell commands via sudo, manage iptables, and run commands in Docker containers (SKILL.md).
Sanitization: There is no evidence of sanitization or filtering of the log data before it is presented to the agent.

debug-firewall