The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses elevated privileges via sudo to perform system-level tasks including package installation (apache2, samba, snmpd), service lifecycle management (systemctl), and writing to system-protected directories.
[COMMAND_EXECUTION]: The workflow includes commands to modify host firewall configurations on both Linux (ufw) and Windows (New-NetFirewallRule) to allow inbound traffic on multiple ports.
[COMMAND_EXECUTION]: Setup instructions for SMB shares include the use of chmod 777 to create globally writable and readable directories, which is an insecure practice outside of isolated lab environments.
[CREDENTIALS_UNSAFE]: The configuration examples for SNMP include the use of well-known and insecure community strings (public, private) for read-only and read-write access.
[PROMPT_INJECTION]: The skill contains a vulnerability surface for indirect prompt injection through log analysis tasks.
Ingestion points: The agent is instructed to read and parse Apache access and error logs (/var/log/apache2/access.log), which contain data provided by external, potentially untrusted network actors (e.g., User-Agent strings or POST request bodies).
Boundary markers: Absent; the skill lacks instructions to treat log data as untrusted or to ignore embedded instructions.
Capability inventory: The skill possesses extensive system capabilities including sudo access, software management, and firewall modification as documented in SKILL.md.
Sanitization: Absent; the provided log parsing commands (grep, awk) do not validate or sanitize log entries before the data is processed by the agent.

Network 101