The Agent Skills Directory

[SAFE]: The skill is purely informational and provides best practices for securing AI agents against prompt injection and social engineering. It focuses on methodology such as defense-in-depth, content trust levels, and immutability of system rules.
[SAFE]: No malicious patterns, obfuscation, or remote code execution were found in the instructions. The skill describes past cases and execution steps to improve security posture rather than performing any operational tasks.
[SAFE]: The content mentions external services (Claude, ChatGPT, Grok) for context and illustrative purposes but does not involve network operations, sensitive file access, or remote script downloads.

injection-defense