foundations-statistical-inference

Pass

Audited by Gen Agent Trust Hub on Sep 23, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The test script scripts/test_split_conformal_quantile.py utilizes subprocess.run to execute the bundled scripts/split_conformal_quantile.py logic. This execution is limited to the local file system and uses the current Python interpreter to verify the correctness of the statistical calculations.
  • [EXTERNAL_DOWNLOADS]: The skill references authoritative statistical resources and research papers from reputable domains including amstat.org and arxiv.org. These links are used for scientific documentation and represent standard academic citations.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided numeric data through a Python helper script.
  • Ingestion points: Data enters via sys.stdin in scripts/split_conformal_quantile.py as a JSON payload.
  • Boundary markers: The script enforces a strict JSON structure requiring specific keys (scores, alpha) and rejects unexpected keys or trailing content.
  • Capability inventory: The skill uses subprocess.run for testing and sys.stdout/stderr for output; no network or sensitive file system write capabilities are present in the processing logic.
  • Sanitization: The implementation uses Decimal and Fraction for exact arithmetic and performs rigorous type checking, ensuring input data is strictly numeric and within valid statistical ranges (0 < alpha < 1).
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 23, 2026, 06:07 PM
Security Audit — agent-trust-hub — foundations-statistical-inference