data-visualization
Pass
Audited by Gen Agent Trust Hub on Mar 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the
infshCLI, a vendor-specific tool for authentication (infsh login) and application execution. - [REMOTE_CODE_EXECUTION]: The skill executes Python and HTML code blocks via remote executors (
infsh/python-executorandinfsh/html-to-image). The provided Python scripts perform local file writes to save charts as PNG files (e.g.,revenue.png,line-chart.png). - [PROMPT_INJECTION]: The skill provides templates for generating code based on data, creating a surface for indirect prompt injection. If the agent populates these templates with unsanitized external data, it could lead to arbitrary code execution within the executor environment.
- Ingestion points: Untrusted data processed for visualization based on triggers like 'chart' and 'graph' in
SKILL.md. - Boundary markers: Absent; the Python recipes use direct value assignment without escaping or delimiters.
- Capability inventory: File system access (writing images) and general Python execution capabilities via
infsh/python-executorinSKILL.md. - Sanitization: No validation or escaping is demonstrated in the provided code snippets.
Audit Metadata