The Agent Skills Directory

[SAFE]: The skill consists entirely of instructional documentation and does not include any executable scripts or commands.\n- [CREDENTIALS_SAFE]: The documentation explicitly advises against storing credentials in experiment records and recommends using configured security mechanisms, following industry best practices.\n- [EXTERNAL_DOWNLOADS]: The skill includes references to reputable academic sources (arXiv) and open-source research repositories (Google, K-Dense-AI) for informational purposes. No automated remote code execution or package installations are performed.\n- [INDIRECT_PROMPT_INJECTION]: While the skill involves processing external data (training datasets), it includes extensive guidance on data validation, semantic review, and structural checks to mitigate risks associated with untrusted data inputs.

ml-experiment