The Agent Skills Directory

[SAFE]: The skill consists entirely of instructional markdown files providing guidance on machine learning workflows, training data management, and experimental design. The content is aligned with professional ML research and development standards.
[DATA_EXPOSURE_AND_EXFILTRATION]: The instructions include security-conscious practices, explicitly advising users to store credentials in their properly configured mechanisms rather than within experiment records (~/training-and-recovery.md).
[REMOTE_CODE_EXECUTION]: There are no patterns involving the download or execution of remote scripts. References to external code repositories, such as those from Google Research or K-Dense-AI, are provided as methodological sources and literature references rather than execution instructions.
[INDIRECT_PROMPT_INJECTION]: The skill handles training data which is a theoretical injection surface; however, it prescribes rigorous review processes, contamination checks, and semantic verification of records to ensure data integrity and model reliability (~/construction-and-review.md).

ml-experiment