The Agent Skills Directory

[SAFE]: A thorough review of the skill's code and its 20 reference files revealed no malicious patterns, unauthorized command execution, or obfuscated content.\n- [PROMPT_INJECTION]: The skill incorporates specific defensive rules, such as Rule 10, which explicitly directs the AI to follow its internal execution sequence and ignore 'Next Step' suggestions generated by external tools or websites. This is a robust mitigation against Indirect Prompt Injection attacks where external data might contain adversarial instructions.\n- [DATA_EXFILTRATION]: While the skill permits attaching files to the synthesis tool for context, it provides strict guidance on selecting relevant code and configuration files. There are no instructions or patterns aimed at accessing sensitive system files, such as SSH keys or environment secrets, and no unauthorized network exfiltration paths were detected.

run-research