research-platform-engineering
Pass
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill scripts and instructions are well-defined and perform only the intended statistical analysis. There is no evidence of obfuscation, remote code execution, or credential theft.
- [PROMPT_INJECTION]: The skill processes external data from a CSV file, which represents a surface for indirect prompt injection.
- Ingestion points:
scripts/research_platform_engineering_diagnostics.pyreads untrusted data frominput.csv. - Boundary markers: No explicit instruction delimiters are provided.
- Capability inventory: The agent executes a Python script to calculate statistics and subsequently reads the resulting JSON report.
- Sanitization: Data is sanitized by converting columns to numeric and datetime types using pandas, which prevents the execution of embedded text instructions within the data fields.
Audit Metadata