The Agent Skills Directory

[SAFE]: The skill consists entirely of educational markdown content. It provides a framework for securing AI agents, including permission scoping, human-in-the-loop controls, and cost bounding. No security risks were identified within the instructional text.
[NO_CODE]: There are no scripts, binaries, or active tool configurations provided with this skill. It functions solely as a process guide.
[PROMPT_INJECTION]: While the document discusses prompt injection as a threat to be mitigated (Step 3), it does not contain any malicious patterns or attempts to override agent behavior. Instead, it correctly identifies retrieved content as a primary attack vector and recommends defensive architectural boundaries.

agent-hardening