build-custom-charts
Pass
Audited by Gen Agent Trust Hub on Jul 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to create and modify JavaScript files within the
agent/charts/directory. These files are subsequently loaded and executed in the browser environment. - [PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by processing results from
execute_sqlqueries. If the data contains malicious scripts, they could be executed in the browser context (XSS) if not properly handled. - Ingestion points: Data object passed to the
renderfunction inSKILL.md. - Boundary markers: No specific markers or instructions are provided to delimit or sanitize untrusted data from instructions.
- Capability inventory: Writing to the project's filesystem and executing arbitrary JavaScript in the user's browser.
- Sanitization: The instructions do not mandate sanitization or validation of the input data, although the example demonstrates using
textContentfor safety.
Audit Metadata