skills/quantskills/skill-intraday-data-quality-auditor/intraday-data-quality-auditor/Gen Agent Trust Hub
intraday-data-quality-auditor
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The repository includes test and validation scripts (tests/test_audit_bars.py and validation/smoke.py) that call subprocess.run to verify the functionality of the main auditing script. These calls are restricted to internal testing, do not use a shell environment, and invoke the current Python interpreter, posing no security risk to the user environment.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data via CSV files in scripts/audit_bars.py. The risk is mitigated by explicit data validation: (1) Ingestion points: CSV files loaded through csv.DictReader in the load_rows function. (2) Boundary markers: None (structured CSV format). (3) Capability inventory: Limited to reading input CSVs and writing JSON reports to specified paths; no network or administrative capabilities. (4) Sanitization: All price and volume fields are converted to float objects with NaN handling for invalid inputs, and timestamps are parsed using datetime.fromisoformat, preventing arbitrary text from influencing agent logic.
- [DYNAMIC_EXECUTION]: The unit tests use importlib.util to load the audit_bars.py module. This is a standard developer practice for testing local scripts and does not involve executing untrusted or remote code.
Audit Metadata