building-recommendation-systems

Pass

Audited by Gen Agent Trust Hub on Apr 12, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect prompt injection surface identified where the skill processes untrusted user interaction data and item features to guide the generation and execution of machine learning scripts.
  • Ingestion points: User preferences, item features, and interaction data as described in the SKILL.md and assets/README.md.
  • Boundary markers: Absent; there are no instructions or delimiters provided to isolate untrusted data from the prompt context.
  • Capability inventory: File system modification and shell command execution are used to create and run generated Python code.
  • Sanitization: Absent; the skill does not describe any validation or filtering of the input data before it influences code generation.
  • [COMMAND_EXECUTION]: The skill uses broad shell execution capabilities to run data preprocessing and model training scripts. This creates a significant attack surface if the agent's code generation is successfully manipulated via the identified prompt injection vectors.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 12, 2026, 04:04 PM