data-viz-2025
Pass
Audited by Gen Agent Trust Hub on Mar 6, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill provides patterns for 'AI-Enhanced Visualizations' that are vulnerable to Indirect Prompt Injection.
- Ingestion points: Data objects are processed and sent to an LLM via
generateInsightfunctions inSKILL.mdandreferences/data-storytelling.md. - Boundary markers: The suggested prompt templates lack boundary markers (e.g., XML tags or clear delimiters) or explicit instructions for the model to disregard instructions embedded within the data.
- Capability inventory: The skill environment grants the agent permissions to read, write, and execute shell commands (
Bash), which could be leveraged if an injection succeeds. - Sanitization: The implementation examples do not include sanitization or validation of the data values before they are interpolated into the prompt string.
Audit Metadata