build-dashboard-theme

Pass

Audited by Gen Agent Trust Hub on Nov 20, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data to derive design decisions, creating a surface for indirect injection attacks.
  • Ingestion points: As described in SKILL.md, input is accepted from brand names, websites, screenshots, brand guidelines, and mood descriptions.
  • Boundary markers: There are no specific boundary markers or instructions defined to isolate or ignore instructions that might be embedded within the provided brand data or websites.
  • Capability inventory: The skill possesses the capability to write .theme.aml configuration files to the .holistics/library/themes/ directory and update dashboard definitions to apply these themes. It also generates custom CSS blocks.
  • Sanitization: The instructions require the use of a strict AML schema and a "Closed Vocabulary," which limits the structure of the output. However, it lacks specific sanitization for the textual values derived from external inputs.
  • [DYNAMIC_EXECUTION]: The skill dynamically generates CSS content based on user input and derived tokens.
  • The custom_css block allows for the inclusion of @import url(...) statements. This allows the skill to generate code that references and loads resources from external network locations at runtime.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 20, 2026, 09:59 AM