The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or data exfiltration attempts were detected in the skill files.
[COMMAND_EXECUTION]: The script scripts/analyze-data-validation.sh performs environment checks for common utilities like jq, yq, and python3, and uses bc for calculations. These operations are limited to local environment discovery and do not execute untrusted input.
[EXTERNAL_DOWNLOADS]: The skill contains references to well-known data quality tools and libraries (e.g., Great Expectations, Deequ, Soda Core) in its documentation, but does not perform any automated downloads or remote code execution.
[INDIRECT_PROMPT_INJECTION]: The skill defines an attack surface by accepting external data files and schemas for validation. However, as the current implementation is a mock script that does not actually parse or process the content of these files, there is no active vulnerability.
Ingestion points: Files passed via --data, --schema, --rules, and --sources in analyze-data-validation.sh.
Boundary markers: None present in the simulation script.
Capability inventory: The script performs no subprocess calls on data content, file-writes, or network operations.
Sanitization: Not applicable as no data processing occurs.

data-validation