marketing-analyst
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data files which provides a potential surface for indirect prompt injection.
- Ingestion points: Processes data from
channels.json,cohort_data.json, andhistorical.jsonvia dedicated Python scripts. - Boundary markers: No specific delimiters or instructions to ignore embedded commands are used when reading data.
- Capability inventory: The agent executes local scripts (
channel_mix_optimizer.py,cohort_analyzer.py,marketing_forecast_generator.py) to process this data and provides budget/strategy recommendations based on outputs. - Sanitization: Input data is parsed as JSON and processed mathematically; however, text fields (e.g., channel names) are not sanitized before being reflected in reports.
- [EXTERNAL_DOWNLOADS]: The skill requires standard third-party libraries for complex statistical calculations.
- Evidence: Code snippets in
SKILL.mdutilizescipyandnumpyfor A/B testing and performance modeling.
Audit Metadata