build-dashboard

Pass

Audited by Gen Agent Trust Hub on Nov 20, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it ingests dataset metadata (field names, descriptions) and external templates to generate dashboard code. However, the risk is mitigated by the skill's requirement for a 'Propose and confirm' workflow, where the AI must wait for explicit user approval before running queries or editing files.
  • [COMMAND_EXECUTION]: The skill uses internal tool calls like fetch_dataset and read_file to understand the project structure and data schema. These operations are scoped to the Holistics project environment and are used for legitimate configuration discovery.
  • [DYNAMIC_EXECUTION]: The skill generates visualization logic in AML. While this is a form of code generation, AML is a domain-specific configuration language for data modeling and visualization, rather than a general-purpose execution environment. The use of HTMLLayout allows for custom HTML/CSS, which is a documented feature of the platform for creating custom reports.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 20, 2026, 09:59 AM