oma-design
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill features a design extraction phase that fetches and analyzes HTML and CSS from user-provided URLs or external projects, creating a vector for untrusted data to enter the agent's context.
- Ingestion points: External reference site source code and project metadata processed during the 'EXTRACT' phase (Phase 2).
- Boundary markers: The execution protocol lacks explicit instructions to use delimiters or ignore embedded instructions when analyzing external source code.
- Capability inventory: The skill is capable of generating project-wide configuration files (DESIGN.md, .design-context.md) and suggests the execution of shell commands for project setup.
- Sanitization: There is no evidence of sanitization, filtering, or validation of the fetched external content before it is processed by the agent.
- [EXTERNAL_DOWNLOADS]: The skill references and suggests the installation of various external packages and CLI tools, including the 'shadcn' library and the '@_davideast/stitch-mcp' package for integration with Google Stitch services.
- [COMMAND_EXECUTION]: The documentation provides several shell commands for project initialization, component installation, and environment setup, such as 'npx shadcn@latest init' and commands to configure the Stitch MCP proxy.
Audit Metadata