data-engineering-data-driven-feature
Pass
Audited by Gen Agent Trust Hub on Feb 27, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it utilizes the
$ARGUMENTSvariable as a direct input for specialized sub-agents across all phases of the development lifecycle. - Ingestion points: The
$ARGUMENTSvariable is processed as raw input in 16 separate steps within theSKILL.mdfile, including those for data analysis, architecture planning, and code implementation. - Boundary markers: There are no delimiters or instructions used to encapsulate user input, increasing the risk that malicious data in
$ARGUMENTScould override the intended task instructions for the sub-agents. - Capability inventory: The sub-agents involved in this workflow possess significant capabilities, including the generation of backend and frontend code, defining data pipeline architectures, and configuring deployment infrastructure (feature flags and rollout rules).
- Sanitization: The skill does not implement any mechanisms to escape, validate, or sanitize the user-provided content before it is interpreted by the language model.
- [COMMAND_EXECUTION]: The workflow explicitly directs sub-agents to generate and implement executable code for backend, frontend, and machine learning components. While this is the intended purpose of the skill, it creates a surface for dynamic code execution (Category 10) where malicious instructions injected into the initial prompt could result in the generation of compromised or backdoored software components.
Audit Metadata