The Agent Skills Directory

[SAFE]: No security issues or malicious patterns were detected. The skill follows best practices by using only the Python standard library and requiring no third-party dependencies or external API access.
[COMMAND_EXECUTION]: The unit tests in scripts/tests/test_check_data_quality.py use subprocess.run to validate the script's command-line interface. These calls are benign as they invoke the skill's own local script using a list of arguments (avoiding shell=True) for testing purposes only.
[INDIRECT_PROMPT_INJECTION]: The skill identifies and processes untrusted markdown files for validation. While this is an ingestion point for external data, the risk is mitigated as follows:
Ingestion points: The scripts/check_data_quality.py script reads the file provided via the --file argument.
Boundary markers: The script does not use explicit boundary markers for LLM prompts; instead, it uses a deterministic regex-based parser to extract financial data.
Capability inventory: The script is limited to reading local markdown files and writing reports to a specified output directory. It has no network access, no administrative privileges, and no capability to execute content found within the documents.
Sanitization: The script performs character normalization (e.g., full-width character conversion) to improve parsing accuracy but does not interpret or execute the input content.

data-quality-checker