The Agent Skills Directory

[OBFUSCATION]: The skill uses a homoglyph domain example (аpple.com) in rules/security-settings.md and AGENTS.md specifically to illustrate the NextDNS IDN Homographs protection feature. This is an educational use case and does not represent a deceptive attack.
[PRIVILEGE_ESCALATION]: Documentation in rules/rewrites.md includes system commands for flushing DNS caches that require sudo privileges. These are standard troubleshooting instructions intended for manual execution by a user, not autonomous exploitation by the agent.
[EXTERNAL_DOWNLOADS]: The skill references the eventsource Node.js package in rules/logs-streaming.md for handling real-time log updates via Server-Sent Events. This is a well-known and legitimate library.
[INDIRECT_PROMPT_INJECTION]: As the skill handles DNS log data, it possesses an inherent attack surface for indirect prompt injection if malicious domain names are stored in logs and subsequently summarized by the agent. This is a generic risk factor for log-processing tools rather than a specific vulnerability in the skill's logic.

nextdns-api