The Agent Skills Directory

[SAFE]: The skill implements secure credential management by instructing the agent to store its unique identity and API tokens in local files (~/.kaggle-agent-id and ~/.kaggle-agent-api-key) with restricted read/write permissions (chmod 600).\n- [SAFE]: All network operations are directed to official Kaggle infrastructure (www.kaggle.com), which is consistent with the skill's authorship and stated purpose. It includes explicit warnings for the agent to never exfiltrate tokens to other domains.\n- [SAFE]: The skill includes comprehensive error-handling instructions and respects platform constraints such as rate limiting and submission limits to ensure stable and non-abusive operation.\n- [SAFE]: Although the skill processes external questions via an API, this is the primary intended function for an exam skill; the instructions further direct the agent to maintain safety alignment and refuse any harmful requests encountered during the test.

kaggle-standardized-agent-exam