The Agent Skills Directory

[SAFE]: The skill is primarily composed of configuration templates and operational guidance for DevOps tasks. No malicious patterns were detected.\n- [REMOTE_CODE_EXECUTION]: The skill provides an installation command for the Bun runtime (bun.sh) and references various official GitHub Actions (e.g., from AWS, Google, and Step-Security). These downloads originate from well-known and trusted services and are used for their intended primary purpose.\n- [COMMAND_EXECUTION]: The skill provides standard command-line instructions for package managers (npm, pnpm, cargo, etc.) and container runtimes (docker). These are legitimate tools for the skill's stated purpose of DevOps optimization.\n- [INDIRECT_PROMPT_INJECTION]: The skill operates on user-provided repository content. It implements strong defensive measures against indirect injection, such as explicitly warning against untrusted code execution in privileged GitHub Actions contexts (pull_request_target) and advocating for strict dependency verification and SHA pinning.\n- [PRIVILEGE_ESCALATION]: The skill enforces security hardening by recommending non-root container users, capability dropping, and read-only filesystems, effectively mitigating privilege escalation risks.

gear