The Agent Skills Directory

[NO_CODE] (SAFE): The skill is composed entirely of markdown documentation. It does not contain any scripts, binaries, or automated command execution patterns.
[PROMPT_INJECTION] (SAFE): No patterns of instruction override, jailbreak attempts, or system prompt extraction were detected. The instructions are focused on providing a framework for code review.
[Category 8: Indirect Prompt Injection] (INFO): This skill is designed to analyze external data (Dockerfiles). While processing untrusted code is a surface for indirect injection, this specific skill lacks any 'write' or 'execute' capabilities (no subprocess calls, no network requests, no file system modifications), meaning it cannot be used as a vector for side effects.
[Category 5: Privilege Escalation] (SAFE): The content explicitly discourages privilege escalation by recommending the USER directive to run as non-root and advising against the --privileged flag.
[EXTERNAL_DOWNLOADS] (SAFE): The example Dockerfile references official images like node:20-alpine, which are from a trusted source (Docker Hub official images).

docker