lw-firewall-management-abilities

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill documents an attack surface where untrusted data (JSON configuration imports and log entries containing visitor-controlled User-Agents/IPs) is processed by the agent.
  • Ingestion points: Data enters the context via Options::save() (JSON import), lw_firewall_log (visitor metadata), and Site Manager abilities (block-ip).
  • Boundary markers: The instructions explicitly warn that "Imported JSON is still untrusted input" and require manual review of IP/CIDR values and email recipients.
  • Capability inventory: The skill can execute shell commands via WP-CLI, install worker files (wp lw-firewall worker install), and modify site access policies (IP banning).
  • Sanitization: The skill notes that OptionSchema::apply() allowlists and clamps values, and IpMatcher::ip_in_cidr() validates CIDR masks.
  • [COMMAND_EXECUTION]: The skill provides detailed mappings for executing system-level commands through the wp-cli interface (e.g., wp lw-firewall config set, wp lw-firewall ban remove, wp lw-firewall worker install). These commands are standard for plugin administration but grant significant control over the host environment and web server configuration.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 07:06 PM
Security Audit — agent-trust-hub — lw-firewall-management-abilities