The Agent Skills Directory

[SAFE]: The skill provides documentation and illustrative code snippets for data cleaning, deduplication, and quality assessment. All operations described are standard machine learning practices.
[REMOTE_CODE_EXECUTION]: No remote code execution or suspicious package installation patterns were detected. The Python snippets reference the cleanlab library, which is a well-known tool for identifying label issues in datasets.
[DATA_EXFILTRATION]: No network operations (e.g., curl, wget, fetch) or access to sensitive local file paths (e.g., SSH keys, environment variables) were found.
[PROMPT_INJECTION]: The instructions do not contain any patterns attempting to override agent behavior or bypass safety guidelines.

Dataset Curator