brand-content-design
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads project-specific identity files (
brand-philosophy.md) and user-provided outlines to influence its design output, creating a surface for indirect prompt injection.\n - Ingestion points: Project root
brand-philosophy.md, template files, and content outlines.\n - Boundary markers: Relies on functional instructions in
references/bias-prevention.md(e.g., 'Reference files are DECISION FRAMEWORKS, not LOOKUP TABLES') rather than technical delimiters.\n - Capability inventory: Scoped to directory searches via
Bash, file creation viaWrite, and invocation of specialized skills for PDF and PPTX generation.\n - Sanitization: Includes automated checks for WCAG contrast and design-value validation to sanitize visual parameters.\n- [COMMAND_EXECUTION]: The skill employs non-privileged shell commands for project management and terminal display. It uses
findto locate project root directories andechowith ANSI escape sequences to preview color palettes in the terminal. These operations are functional and restricted to the local environment.\n- [EXTERNAL_DOWNLOADS]: A reference URL for design token methodology (https://camoa.github.io/dev-guides/...) is included in the palette documentation. This link points to the vendor's own documentation site and does not involve the execution of remote scripts or code.
Audit Metadata