experiment-designer

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill instructions and referenced documents contain no malicious patterns, prompt injections, or unauthorized network operations.
  • [COMMAND_EXECUTION]: The skill includes instructions for running a local Python utility (scripts/sample_size_calculator.py) to estimate required sample sizes. The script uses standard mathematical modules (math, statistics) and performs no network, file system, or dynamic code execution tasks.
  • [INDIRECT_PROMPT_INJECTION]: The skill involves processing user-supplied numeric parameters for statistical analysis. 1. Ingestion points: Numeric arguments to the sample size script. 2. Boundary markers: Parameter values are constrained by command-line argument parsing. 3. Capability inventory: The script contains no capabilities for network access, file writing, or subprocess spawning. 4. Sanitization: The script performs range checks to ensure conversion rates and significance levels are mathematically valid (between 0 and 1).
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 05:13 PM
Security Audit — agent-trust-hub — experiment-designer