The Agent Skills Directory

[SAFE]: The skill is focused on privacy engineering and data anonymization. All code components in SKILL.md and scripts/process.py perform legitimate statistical operations, data modeling, and policy validation without attempting any unauthorized actions.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface as it processes external DataFrames and distribution metrics to generate synthetic data and quality reports. However, the risk is negligible as the skill does not have access to tools that could be exploited (e.g., network access or shell execution).
Ingestion points: DataFrames passed to SDV synthesizer training functions in SKILL.md and distribution dictionaries passed to SyntheticDataAssessor in scripts/process.py.
Boundary markers: None identified in the provided scripts.
Capability inventory: The skill is limited to mathematical and statistical operations using pandas, numpy, and sdv; it does not utilize subprocesses, network requests, or file-writing tools.
Sanitization: No specific input sanitization is present, which is typical for statistical processing libraries.

privacy-data-sharing