The Agent Skills Directory

[SAFE]: The skill consists entirely of markdown-based security guidelines and instructions. It does not include any executable scripts, automated installations, or network-active components.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted code and diffs during reviews (SKILL.md). While this presents an ingestion surface for potentially malicious data, the skill's primary function is to provide the agent with the necessary boundary markers and validation logic to neutralize such threats. Furthermore, the skill itself lacks any dangerous capabilities (subprocess execution, file writes, or network operations) that could be exploited via injection.
[CREDENTIALS_SAFE]: The skill provides explicit instructions to prevent the exposure of credentials, such as hardcoding secrets or leaking them in logs and client bundles (references/secret-handling.md). It does not contain any hardcoded secrets itself.
[EXTERNAL_DOWNLOADS]: The skill references trusted external resources, including official IETF RFCs and GitHub documentation, which are used purely for educational reference. No remote code execution or unvetted package downloads are performed.

application-security