data-analysis
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data inputs which can serve as a vector for indirect prompt injection. A malicious dataset could contain hidden instructions aimed at overriding the agent's analysis logic or statistical gates.
- Ingestion points: Datasets are loaded in Phase 3 (EXTRACT) as specified in SKILL.md.
- Boundary markers: The framework does not incorporate specific delimiters or 'ignore' instructions for data content.
- Capability inventory: The skill utilizes Bash for execution and file manipulation tools (Edit, Write).
- Sanitization: No data-cleansing or instruction-filtering procedures are defined for the extracted data.
- [DYNAMIC_EXECUTION]: The workflow involves the creation and execution of transient Python scripts for metric computation and data profiling.
- Evidence: Instructions in SKILL.md and code patterns in references/compute-examples.md guide the agent to generate and run analysis code dynamically.
Audit Metadata