The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to perform various system operations using powerful command-line utilities to manage firewall states and persistence.
Recommended commands include nft (with flags for check, apply, and list), systemctl (for status, cat, and dependency analysis), journalctl (for log monitoring), and ripgrep (rg) for recursive system directory searches.
Evidence: Found throughout operational guides, specifically in references/nftables-ops-cheatsheet.md and references/debug-workflow.md.
[PROMPT_INJECTION]: The skill exhibits a vulnerability surface for indirect prompt injection because its core workflow involves the ingestion and analysis of external data that may contain malicious instructions.
Ingestion points: The agent is tasked with reviewing, linting, and debugging ruleset snippets or .nft files provided by users or other automated agents (identified in SKILL.md and references/debug-workflow.md).
Boundary markers: The provided instructions do not specify the use of delimiters or specific warning prompts to prevent the agent from being influenced by instructions embedded within the firewall rulesets under review.
Capability inventory: The skill is equipped with capabilities to execute system commands (nft -f), read service configurations (systemctl cat), and perform broad file system searches (rg), which could be misdirected by a successful injection (documented in references/nftables-ops-cheatsheet.md).
Sanitization: There is no evidence of automated validation, sanitization, or filtering of the input ruleset content before the agent processes and acts upon it.

nftables-rule-writing