The Agent Skills Directory

[PROMPT_INJECTION]: No patterns of instruction override, jailbreak attempts, or safety filter bypasses were detected. The instructions are strictly limited to the stated purpose of resource management.- [DATA_EXFILTRATION]: No access to sensitive local files (e.g., SSH keys, credentials) or network transmission of data was found.- [REMOTE_CODE_EXECUTION]: No remote scripts, package installations, or dynamic code execution logic are included in the skill.- [CREDENTIALS_UNSAFE]: No hardcoded secrets, API keys, or private tokens were identified in the content.- [SAFE]: No obfuscation (Base64, zero-width chars) or persistence mechanisms were discovered during analysis.- [NO_CODE]: The skill does not bundle any scripts, binaries, or tool configurations, significantly reducing the executable attack surface.

bench-management