stitch-design-md
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a provided shell script
scripts/fetch-stitch.shto download design assets. This script usescurlto fetch remote resources andmkdirto manage local directories, which are standard operations for its primary purpose. - [EXTERNAL_DOWNLOADS]: The skill is designed to fetch HTML content from Google Cloud Storage using URLs provided by the Stitch MCP server. These downloads are required to retrieve the necessary source code for design analysis.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection through its ingestion of external data.
- Ingestion points: Design assets are downloaded to
temp/source.htmland parsed by the agent to extract Tailwind classes and design tokens. - Boundary markers: No specific delimiters or instructions to ignore embedded commands are included in the analysis prompts.
- Capability inventory: The skill has permissions to execute local scripts, perform file system operations, and interact with the Stitch MCP tools.
- Sanitization: The skill does not perform any sanitization or validation of the downloaded HTML content before the agent analyzes it.
Audit Metadata