The Agent Skills Directory

[SAFE]: The skill is composed entirely of reference documentation and instructional markdown files. It contains no executable scripts, binaries, or automated shell commands. All technical rules are presented as textual guidance for the agent to apply when interacting with networking tools.
[SAFE]: All external URLs provided in the inventory and documentation point to official documentation sites, RFC repositories, and well-known technology organizations (e.g., Cloudflare, Docker, Microsoft, Netgate). No suspicious domains or typosquatting attempts were found.
[INDIRECT_PROMPT_INJECTION]: The skill defines a surface for processing untrusted data by instructing the agent to review and write network configurations provided by users. However, the skill actively mitigates risks by teaching the agent how to identify and secure common vulnerabilities, such as header smuggling in proxies or cryptokey routing in VPNs.
Ingestion points: Network configuration files (nftables, Caddyfile, Traefik config, etc.) that the user asks the agent to review.
Boundary markers: The reference material does not explicitly define delimiters for untrusted data, relying on the agent's ability to interpret standard configuration syntax.
Capability inventory: Analysis and creation of networking files. The skill files themselves do not invoke any tools or make network calls.
Sanitization: The skill provides extensive rules for verifying the integrity of configurations (e.g., ensuring identity headers are stripped from proxies), which serves as a form of semantic sanitization during the review process.

networking