dummy-dataset
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided inputs through arguments such as
$COLUMNSand$CONSTRAINTSto define data generation logic. - Ingestion points: Arguments passed to the instructions in
SKILL.mdused to define the dataset structure. - Boundary markers: None present; the instructions do not include delimiters or warnings to ignore embedded instructions in the user-supplied constraints.
- Capability inventory: Generates Python scripts and data files (CSV, JSON, SQL). No direct network or privilege escalation capabilities are present in the provided template.
- Sanitization: None present; the skill relies on the agent to interpret inputs directly.
- [DYNAMIC_EXECUTION]: The skill generates Python scripts at runtime based on user-supplied parameters to create the datasets.
- Evidence: The
SKILL.mdfile contains a Python template and instructions to "Create executable scripts... for immediate use." - Assessment: The generated script uses standard library modules (csv, json, random) and follows a static template, posing a low risk if execution is limited to the defined logic.
Audit Metadata